Brennan, Niamh M., Guillamon-Saorin, Encarna and Pierce, Aileen [2009] Impression Management: Developing and Illustrating A Scheme of Analysis for Narrative Disclosures – A Methodological Note. Accounting, Auditing and Accountability Journal, 22(5): 789

Uploaded on

Purpose – This paper develops a holistic measure for analysing impression management and for detecting bias introduced into corporate narratives as a result of impression …

Purpose – This paper develops a holistic measure for analysing impression management and for detecting bias introduced into corporate narratives as a result of impression management.
Design/methodology/approach – Prior research on the seven impression management methods in the literature is summarised. Four of the less-researched methods are described in detail, and are illustrated with examples from UK Annual Results’ Press Releases (ARPRs). A method of computing a holistic composite impression management score based on these four impression management methods is developed, based on both quantitative and qualitative data in corporate narrative disclosures. An impression management bias score is devised to capture the extent to which impression management introduces bias into corporate narratives. An example of the application of the composite impression management score and impression management bias score methodology is provided.
Findings – While not amounting to systematic evidence, the 21 illustrative examples suggest that impression management is pervasive in corporate financial communications using multiple impression management methods, such that positive information is exaggerated, while negative information is either ignored or is underplayed.
Originality/value – Four impression management methods are described in detail, illustrated by 21 examples. These four methods are examined together. New impression management methods are studied in this paper for the first time. This paper extends prior impression management measures in two ways. First, a composite impression management score based on four impression management techniques is articulated. Second, the composite impression management score methodology is extended to capture a measure for bias, in the form of an impression management bias score. This is the first time outside the US that narrative disclosures in press releases have been studied.

  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
    Be the first to like this
No Downloads


Total Views
On Slideshare
From Embeds
Number of Embeds



Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

    No notes for slide


  • 1. Impression management: developing and illustrating a scheme of analysis for narrative disclosures – a methodological note Niamh M. Brennan#, Encarna Guillamon-Saorin* and Aileen Pierce# # University College Dublin *Universidad Carlos III de Madrid (Published in Accounting, Auditing and Accountability Journal, 22(5)(2009): 789-832.)Address for correspondence:Prof. Niamh Brennan, Quinn School of Business, University College Dublin, Belfield, Dublin 4. Tel. +353-1-716 4704; Fax +353-1-716 4767; e-mail Niamh.Brennan@ucd.ieAcknowledgementsThis research was partially supported by the Accounting Harmonisation and Standardisation in Europe:Enforcement, Comparability and Capital Market Effects research project (Contract No. HPRN-CT-2000-00062) carried out by the HARMONIA network and funded by the European Commission ResearchTraining Programme.The authors are grateful to two anonymous referees for their very helpful advice.
  • 2. Impression management: developing and illustrating a scheme of analysis for narrative disclosures – a methodological noteAbstractPurpose – This paper develops a holistic measure for analysing impressionmanagement and for detecting bias introduced into corporate narratives as a result ofimpression management.Design/methodology/approach – Prior research on the seven impressionmanagement methods in the literature is summarised. Four of the less-researchedmethods are described in detail, and are illustrated with examples from UK AnnualResults’ Press Releases (ARPRs). A method of computing a holistic compositeimpression management score based on these four impression management methodsis developed, based on both quantitative and qualitative data in corporate narrativedisclosures. An impression management bias score is devised to capture the extent towhich impression management introduces bias into corporate narratives. An exampleof the application of the composite impression management score and impressionmanagement bias score methodology is provided.Findings – While not amounting to systematic evidence, the 21 illustrative examplessuggest that impression management is pervasive in corporate financialcommunications using multiple impression management methods, such that positiveinformation is exaggerated, while negative information is either ignored or isunderplayed.Originality/value – Four impression management methods are described in detail,illustrated by 21 examples. These four methods are examined together. Newimpression management methods are studied in this paper for the first time. Thispaper extends prior impression management measures in two ways. First, a compositeimpression management score based on four impression management techniques isarticulated. Second, the composite impression management score methodology isextended to capture a measure for bias, in the form of an impression management biasscore. This is the first time outside the US that narrative disclosures in press releaseshave been studied.Keywords Impression management, press releases, content analysisPaper type Methodology paper ii
  • 3. 1. IntroductionWhat is impression management?Impression management has its origins in the psychology literature (Schlenker, 1980;Riess et al., 1981; Schneider, 1981). The term “impression management” refers to theprocess by which individuals attempt to control the impressions of others (Leary andKowalski, 1990, p. 34). In the context of corporate reporting, impression managementoccurs when management selects information to display and presents that informationin a manner that distorts readers’ perceptions of corporate achievements (Neu, 1991;Neu et al., 1998). Impression management predominantly occurs in less regulatednarrative disclosures which focus on interpreting financial outcomes.Exercise of discretion and managerial motives for impression managementMost accounting studies of impression management are based explicitly or implicitlyon the assumption that management is motivated by a desire to present a self-servingview of corporate performance (Neu, 1991; Neu et al., 1998). This manifests itself ina number of ways. Firstly, management is hypothesised to want to hide poor firmperformance. (Adelberg, 1979) suggested that managers might be expected toobfuscate their failures and underscore their successes. The obfuscation hypothesis,first tested by Courtis (1995), posits that management is not neutral in how it presentsinformation, preferring to communicate in a manner that hides bad news. Forexample, text reporting negative organisational outcomes is expected to use languageand syntactical features that make the text more difficult to read. Management mayuse rhetorical devices to conceal negative organisational outcomes. A variant on thisrelates to attributional behaviour by management. In narrative explanations ofperformance, it is assumed management will act in a self-interested manner andattribute poor performance to external factors or to other factors outside its control(e.g., predecessor CEOs), and attribute good performance to internal factors (i.e., theirown good management). Management is expected to manipulate themes by disclosingmore positive and less negative information in the form of accounting narratives. Avariation on this is selection by management of quantitative amounts such as earningsnumbers for disclosure that display the firm in the best possible manner. Visual andpresentation techniques are also expected to underplay negative performance andexaggerate positive performance (see, for example, Beattie and Jones, 2002; Courtis,2004a). Finally for the purpose of showing management in the best possible light, 1
  • 4. accounting narratives are expected to contain performance comparators andbenchmarks that display the company most favourably.Objectives and contribution of the paperBeattie et al. (2004, p. 213) observe “…extant approaches to the analysis ofaccounting narratives…are essentially one dimensional, whereas disclosure is acomplex multi-faceted concept”. The purpose of this paper is to develop a holisticmeasure of impression management for both qualitative and quantitative disclosures.This paper has four objectives: (1) The impression management literature is reviewedfrom a methodological perspective, summarising the seven methods examined in priorresearch to measure impression management; (2) Four of the less-researched methodsof measuring impression management are described in depth and are illustrated usingexamples from annual results press releases (ARPRs) of UK companies; (3) A holisticmethod of measuring impression management is developed in the form of compositeimpression management scores based on the four impression management techniquesstudied in the paper. A composite impression management score for both qualitativeand quantitative disclosures is articulated. Finally, based on the composite impressionmanagement score methodology, (4) a method for measuring the bias introduced intonarrative disclosures by impression management is developed, in the form of animpression management bias score.The paper restricts itself to narrative disclosures. This study focuses on press releasesannouncing annual results. These press releases contain information on companyperformance, thereby facilitating an analysis of the influence of performance onimpression management. Press releases are voluntary disclosures, released bycompanies to the market (i.e., to the media, shareholders, wire services, etc.) eventhough not required by laws or regulations. The content of press releases is largely(but not completely) unregulated, and this makes it easier for managers to manipulatethe information disclosed therein, and a potential vehicle for impression management.Their coverage in national newspapers, television and radio business reports by anoften uncritical financial media, provides them with a wider audience beyond userswho study annual reports (see Maat, 2007 for a more detailed discussion of theseissues). Thus, their influence on user perceptions is arguably far in excess of those ofother accounting disclosure vehicles. 2
  • 5. The paper contributes to the literature in six ways. (1) In-depth insights into fourcontent analysis techniques are provided, illustrating those techniques with 21examples from UK ARPRs. These are among the less-researched impressionmanagement techniques. (2) Most prior research considers a single, or at most two,impression management techniques in a given context. This study applies fourtechniques: thematic form-orientated analysis, selectivity (choice/selection ofperformance number), visual/presentation effects (emphasis) and performancecomparisons (use of benchmarks). (3) New impression management techniques arestudied for the first time in accounting. Visual/presentation effects (emphasis), isarticulated in three different ways (location/positioning, repetition, reinforcement). (4)A holistic composite impression management score for impression management isarticulated, based on four (out of a maximum of seven) impression managementmethods. (5) An impression management bias score is computed to measure theextent to which impression management introduces bias into financial reporting. (6)Nearly all of the existing literature (mainly UK, US and Australia) is based onnarratives in corporate reports (commonly the president’s letter/chairman’sstatement). Basing this study on narratives in press releases represents an opportunityto extend research findings to a new communications format. While there has been asmall number of studies of disclosures in press releases in the US (e.g., Lougee andMarquardt, 2004; Bowen et al., 2005; Johnson and Schwartz, 2005; Davis et al.,2007; Henry, 2008), this is the first study of disclosures in press releases in anothercountry. Press releases are important disclosure vehicles given their subsequentinfluence on, and even inclusion in, media outlets resulting in wider dissemination oftheir content compared with the content of annual reports.Organisation of the paperPrior research on impression management focusing on methodological aspects of thatresearch is reviewed in Section 2. Section 3 sets out in detail the methods applied inthis study. These are illustrated using examples from various press releases in Section4. The paper concludes in Section 5 with a discussion of the implications of thefindings, limitations of the research, and the opportunities for future research. 3
  • 6. 2. Prior impression management researchThis section of the paper commences by considering the wide variety of disclosurevehicles which contain accounting narratives and which provide management withopportunities to manage readers’ impressions. Prior impression management researchis then reviewed from the perspective of the methods used to analyse impressionmanagement tactics and practices.Vehicles for impression managementThe focus of prior research on accounting narratives has been Presidents’ Letters,Chairmen’s Reports, Management Discussion & Analyses, Operating and FinancialReviews, Auditors’ Reports, Financial Statement Footnotes, Interim reports,Prospectuses, Press Releases and Environmental Disclosures (see Tables 2 to 5 inMerkl-Davies and Brennan, 2007).Methods of content analysis in prior impression management researchTable 1 summarises the content analysis methods applied in analysing accountingnarratives from an impression management point of view in prior research. Sevenapproaches have been identified, including syntactic manipulation, rhetoricalmanipulation, attribution of organisational outcomes (meaning-orientated studies),thematic manipulation (form-orientated studies), selectivity (choice/selection ofperformance number), visual/presentation effects (emphasis), and performancecomparisons. The latter four methods are applied in this paper. 4
  • 7. Table 1: Impression management in corporate documents: Content analysis methods in prior research(1) Syntactical manipulation (method of (2) Rhetorical manipulation (method of analysis applied) (4) Thematic manipulation - Form-orientated studies (method ofanalysis applied) • Thomas (1997) (Passive constructions, Sentence openers, Relationship between analysis applied) • Adelberg (1979) (Cloze) first and last paragraph, Euphemisms) • Tennyson et al. (1990) (WORDS) • Parker (1982) (Fog) • Jameson (2000) (Multiple voices, Embedded genres, Contrasting focal points) • Abrahamson and Park (1994) (Negative keywords) • Lewis et al. (1986) (Fog, Flesch, Kwolek, • Sydserff and Weetman (2002)2 (Transitivity index, DICTION) • Smith and Taffler (1995) (User perception) Dale-Chall, Lix, Fry) • Yuthas et al. (2002) (Comprehensibility, Truth, Legitimacy, Sincerity, DICTION) • Abrahamson and Amir (1996) (Negative keywords) • Courtis (1986) (Fog, Flesch) • Smith and Taffler (2000) (Positive/negative keywords) • Jones (1988) (Flesch) (3) Attribution of organisational outcomes - Meaning-orientated thematic studies • Lang and Lundholm (2000) (Type of statements (performance, • Baker and Kare (1992) (Flesch) (method of analysis applied) management spin, forward looking, other), Tone of disclosures • Stevens et al., (1992)1 • Ingram and Frazier (1980)3 (WORDS) (optimistic, pessimistic)) • Smith and Taffler (1992a; 1992b) (Flesch, • Frazier et al. (1984) (WORDS) • Clatworthy and Jones (2003)5 (Positive/negative keywords and Lix, Cloze) • Staw et al. (1983) (Analysis of performance explanations) statements) • Subramanian et al. (1993) (Fog, Flesch) • Aerts (1994) (Analysis of performance explanations) • Rutherford (2005) (Frequencies of 90 keywords) • Courtis (1995) (Fog, Flesch, Lix) • Baginski et al. (2000) (Manual coding of internal / external causes) • Davis et al. (2007) (Optimistic/Pessimistic language use, DICTION) • Jones (1997)1 • Hooghiemstra (2001) (Performance explanations, Technical language) • Henry (2008) (Tone (Frequency positive/negative keywords – • Courtis (1998) (Flesch) • Aerts (2001) (Analysis of performance explanations) DICTION)), Length of press release, Textual complexity, Numerical • Sydserff and Weetman (1999) (Flesch, • Clatworthy and Jones (2003)4 intensity) Texture index) • Lee et al. (2004) (Attributional statements) • Clatworthy and Jones (2006) (Length of accounting narratives, • Clatworthy and Jones (2001) (Flesch) • Baginski et al. (2004) (Manual coding of internal / external causes) Number passive sentences, Number key financial indicators, • Sydserff and Weetman (2002)2 (Flesch, • Aerts (2005) (Analysis of performance explanations) Number personal references, Number quantitative references, Transitivity index, Diction) • Ogden and Clarke (2005)5 Number future references) • Rutherford (2003) (Flesch) • Barton and Mercer (2005) (Analysis of managerial attributions) • Courtis (2004b) (Flesch) • Matsumoto et al. (2006)(Number words positive/negative tone) • Li (2008) • Merkl-Davies (2007) (5) Selectivity (Choice/selection of performance number) • Lougee and Marquardt (2004) (Pro forma earnings disclosures) • Johnson and Schwartz (2005) (Pro forma earnings disclosures) (6) Visual/presentation effects (emphasis) • Staw et al. (1983) (Ordering of information) • Courtis (1996) (Repetition) • Baird and Zelin (2000)(Ordering of information) • So and Smith (2002) (Use of colour) • Courtis (2004a) (Use of colour) • Bowen et al. (2005) (Emphasis/positioning of pro forma earnings) • Kelton (2006) (Design characteristics of the accounting information) • Elliot (2006) (Emphasis/presentation of GAAP and pro forma earnings) (7) Performance comparisons • Lewellen et al. (1996) (Stock return performance comparisons) • Schrand and Walther (2000) (Prior period earnings benchmarks) • Cassar (2001) (Disclosure of share performance graphs) • Short and Palmer (2003) (Performance referents) • Krische (2005)(Prior period benchmark comparisons of earnings) 5
  • 8. 1 Methodological discussion.2 Sydserff and Weetman (2002) is difficult to classify as it uses three methods: one reading ease manipulation and two rhetorical manipulation.3 Ingram and Frazier (1980) is a corporate social reporting study.4 Clatworthy and Jones (2003) test both for the association between positive/negative organizational outcomes and increasing/declining performance and the attribution of positive/negative organizational outcomesto internal/external factors and increasing/declining/performance.5 Odgen and Clarke (2005) examine impression management in the context of legitimacy. They use attribution of organisational outcomes in the form of entitlements and excuses as part of a whole array ofimpression management techniques aimed at gaining legitimacy.Source: Reproduced (with amendments) from Table 2 in Merkl-Davies and Brennan (2007) 6
  • 9. Syntactical manipulation in accounting narrativesThe largest group of studies comprises an analysis of the language used in accountingdisclosures. Much of this research is motivated by the assumption that managers uselanguage to obfuscate corporate performance, especially negative performance. Thehypothesis is that negative performance is reported using language that is more difficultto read.These syntactic studies apply various methods of measuring readability, focusing onanalysing the readability of the text using features such as sentence length or number ofsyllables. Readability is assessed by a readability formula which counts languagevariables in a text in order to provide a measure of probable reading difficulty for readers.As shown in Table 1, ten different readability measures have been applied in prioraccounting research: Fog, Flesch, Kwolek, Dale-Chall, Lix, Fry, Cloze, Texture index,Transitivity index and Diction. Most studies investigating impression managementthrough readability look at the relationship between readability and companyperformance (Adelberg, 1979; Courtis, 1986; Jones, 1988; Baker and Kare, 1992; Kohutand Segars, 1992; Smith and Taffler, 1992a; Subramanian et al., 1993; Courtis, 1995,1998; Clatworthy and Jones, 2001).Readability methodology has been criticized because it originated in the psychologyliterature where it is used to assess children’s writing. Various authors haveacknowledged the limitations of readability formulae and their application to accountingnarratives, giving rise to issues of validity (Jones and Shoemaker, 1994, pp. 164-5;Courtis, 1998).Rhetorical manipulation in accounting narrativesThis stream of impression management research is also based on the obfuscationhypothesis, whereby management makes linguistic choices and uses rhetorical devices toconceal negative firm performance.Rhetorical manipulation involves the exercise of linguistic choices to influence meaning.Rhetoric is defined as “the art of using language so as to persuade or influence others; 7
  • 10. speech or writing expressed in terms calculated to persuade or impress (often in adepreciatory sense), language characterized by artificial or ostentatious expression”(Oxford English Dictionary, 1989, p. 857). Llewellyn (1999) observes that the extent towhich the point of a story is persuasive/convincing/credible depends on its rhetoricalpower, which in turn is a function of linguistic techniques such as plots, labelling,metaphor and platitude. To date there has been little research examining rhetoric andargument in financial reporting (exceptions include Warnock, 1992, 2000; Brennan andGray, 2000). Covaleski et al. (1995, p. 26) comment that “…accounting is not only aninstrument for representing an economic reality…but also a rhetorical device for settingforth…”. Thompson (1991, p. 573) states that “the way theories are justified andlegitimated becomes much more one of a debate, conversation or argument in which theattempt is to persuade an assumed sceptical audience. Hence the interest in rhetoric andin the protocols of argumentation.”Brennan and Gray (2000) examine disclosures in profit forecasts and in takeoverdocuments from the perspective of rhetoric and argument to show how managements useaccounting information to defend their own position and rebut the arguments of the otherside. Persuasion in forecasts, and the verbal jousting and argument between bidder andtarget managements during contested bids, is considered. The plausibility and credibilityof the language used and the arguments offered are analysed.Attribution of organisational outcomes - Meaning-orientated thematic studiesAn alternative to syntactic analysis is thematic analysis. Prior literature using thisapproach forms two distinct groups: meaning-oriented and form-oriented studies (Smithand Taffler, 2000). Form-orientated studies are discussed further on in connection withattribution in narrative financial reporting.Meaning-oriented studies using thematic analysis, summarised in Table 1, investigatepatterns of causal reasoning and attribution used to explain corporate performance. Forexample, Frazier et al. (1984) apply factor analysis to extract themes from managementanalyses of the results of operations in annual reports. The scores from the factor analysisare used to test for differences between good and bad performers and between owner-controlled and management-controlled companies. These studies find that management 8
  • 11. has a tendency towards self-enhancement by attributing responsibility for positiveoutcomes to internal organisational factors; and towards self-protection by attributingresponsibility for negative outcomes to external circumstances.Thematic manipulation in accounting narratives – Form-orientated studiesForm-oriented studies predominantly revolve around the use by management of positiveand negative themes, analysing word and sentence frequencies in order to drawinferences. Examples include Abrahamson and Park (1994), Abrahamson and Amir(1996), Clatworthy and Jones (2003) and Clatworthy and Jones (2006). Similar to someother content analysis techniques, there is a degree of subjectivity involved in thisanalysis, as it relies upon the classification of keywords into positive and negativecategories.For the method to be reliable, it must include a correct measurement specification (Jonesand Shoemaker, 1994). To date, form-oriented content analysis studies investigatingimpression management have been relatively simplistic, for example, using word counts(Clatworthy and Jones, 2003), number of sentences (Kohut and Segars, 1992), coding ofcertain words (Sydserff and Weetman, 2002). Clatworthy and Jones (2006) is morecomprehensive, considering textual characteristics such as quantitative disclosures, keyfinancial performance variables in the text, personal references, passive sentences, future-orientated sentences.SelectivityAlthough selectivity in graphs has been studied extensively in prior research, the scope ofthis paper is restricted to selectivity in narrative disclosures. Given discretion,management may select performance numbers (most commonly earnings numbers) toreport/highlight in narratives that portray firms in the best possible light. Selectivity maybe based on numbers generated from generally accepted accounting principles (GAAP)and non-GAAP numbers (Lougee and Marquardt, 2004). The term “pro forma” earningsis commonly used in respect of earnings numbers other than those calculated underGAAP. Two possible explanations for management use of pro forma earnings have beenput forward (Johnson and Schwartz, 2005): (1) Management are motivated to provideinvestors with more accurate and/or more useful information or (2) Managers deliberately 9
  • 12. make the firm look more profitable. Where the latter motivation exists, use of pro formaearnings fits the definition of impression management.Consistent with agency theory, managers are likely to select the metric that portrays thefirm in the best light (although there are occasions when the reverse might be true, e.g.,big bath accounting). There is widespread evidence that pro forma earnings numbersreported in narrative disclosures in press releases are predominantly income increasingover their GAAP counterpart (Johnson and Schwartz, 2005). This supports an impressionmanagement motivation for such reporting. Johnson and Schwartz (2005, p. 924) refer tousing pro forma earnings for the purpose of “managing readers’ perceptions ofearnings”. They find support for managerial self-serving behaviour in that pro formaearnings exclude more than non-recurring items. They also find that firms that report proforma earnings have earnings that are no different in persistency compared with firmsthat report GAAP earnings. This, they say, contradicts the notion that firms use pro formaearnings to draw investors’ attention to less persistent, more transitory items in GAAPearnings.Visual/presentation effects (emphasis)Jameson (2000, p. 33) observes that discourse has both verbal and visual elements thatinteract with one another. Visuals such as graphic highlighting, headings, bulleted ornumbered lists, colour, shading, logos, may foreshadow verbal discussion or reinforcekey points. There are three different ways of emphasising disclosures in narrativefinancial reporting documents. Firstly, visual emphasis occurs when companies usepresentation techniques to make a piece of information more obvious to readers.Examples of such visual emphasis include locating or positioning of disclosures, oremphasis of text using bullet points, bold text, colour, etc. (So and Smith, 2002; Courtis,2004a). A second form of emphasis is repetition, which occurs when an item is repeated.While Courtis (1996) treats repetition as redundant information, we take a differentapproach interpreting its usage as a form of emphasis. Finally, reinforcement is a form ofemphasis which occurs when a piece of information is emphasised by using a qualifier(an additional word to add emphasis to a keyword, e.g., “Strong growth” – “growth” isthe keyword, “strong” is the qualifier). 10
  • 13. Bowen et al. (2005) examine the way in which pro forma earnings and GAAP earnings inpress releases are emphasised. They measure emphasis in two ways: positioning of thedisclosure item of interest (pro forma earnings; GAAP earnings) in the press release, andthe relative positioning of pro forma compared with GAAP earnings. They find thatmanagers emphasise the metric that portrays the firm in a better light.Impression management using performance comparisonsAnother technique to create an impression of performance that may be biased is to choosebenchmarks which portray current firm performance in the best possible light(performance comparisons). Lewellen et al. (1996), Schrand and Walther (2000), Cassar(2001) and Short and Palmer (2003) investigate the selective use of a benchmark tohighlight positive changes in earnings. Performance comparisons have been studied in thecontext of performance referents, benchmark earnings number, and benchmarkcomparisons in proxy statements and share performance graphs.One of the first benchmark studies was Lewellen et al. (1996) who examined ordinaryshare price performance benchmarks disclosed in corporate proxy statements. They foundthat the benchmarks chosen were biased downwards, which had the effect of allowingmanagement to overstate relative share return performance.Short and Palmer (2003) investigate the way CEOs monitor and interpret organizationalperformance by means of comparisons of performance indicators against internal (such aspast performance) and external (such as competitors and industry averages) referencepoints. They perform content analysis on Presidents’ Letters to Shareholders of 116 UScompanies. They find a strong preference for the use of internal referents (85.4%) ascompared with external referents (14.6%) to assess performance. They find CEOs oflarge and well-performing companies use more external referents (comparisons withcompetitors and industry averages) in their performance explanations than those of smalland poorly-performing companies.Schrand and Walther (2000) find that managers are more likely to select the lowest priorperiod comparative benchmark earnings number that enables them to report the highestyear-on-year increase in earnings. 11
  • 14. Cassar (2001) investigates use of benchmark comparisons in share performance graphs.Almost all sample companies (87%) perform (share price performance and accumulatedshare investment) better than their benchmark (generally market indexes). This suggeststhat, when managers have discretion, they select the information presenting the bestperformance for the company.This review of prior literature has pointed to many different tactics and methods ofimpression management in corporate narrative disclosures in financial reports. Thesemethods have tended to be considered individually in prior studies. Of the seven priorimpression management methods identified in Table 1, thematic manipulation (form-orientated studies), selectivity (choice/selection of performance number),visual/presentation effects (emphasis), and performance comparisons are applied in thisresearch to the analysis of disclosures in press releases. Literature reviewed here suggeststhat these four techniques are among the least researched in prior literature. These fourtechniques lend themselves to the content analysis methodology applied in the research.3. Methods used to measure impression management in this paperThis section discusses the approach taken to analyse ARPRs. As all four methods involvemanual content analysis, the benefits of manual content analysis and computer-aidedapproaches are compared. Examples from press releases are provided to illustrate thedifferent impression management techniques used in ARPRs.Data sourcesThe disclosure vehicle chosen for this study is the ARPR. As previously mentioned, theseare important disclosure vehicles given their wider dissemination in the media. ARPRswere chosen for a number of reasons: (1) Most listed companies issue such a pressrelease. (2) The content of ARPRs is more comparable with respect to content given thatthey all have a common purpose (to announce annual results). (3) Some measures ofimpression management require a performance number (e.g., selectivity, performancecomparisons). Performance numbers are more likely to appear in ARPRs, than pressreleases announcing other corporate events. 12
  • 15. ARPRs were first gathered from official sources (Regulatory News Service-RNS). Wherethe press release was not available from this source, the company website was searched.Where the press release was not available from these public sources, the press release wasobtained directly from the company.Twenty one illustrative examples (excluding Appendix 1) of disclosure practices wereselected from a sample of 101 UK ARPRs. For the purposes of this paper, the examplesselected are ad hoc and serve to provide illustrative rather than systematic evidence ofdisclosure practices. However, as impression management may be influenced bycompany performance, a distinction is made between good news and bad newscompanies. Classification of good/bad news companies is by reference to whetherreported profits were higher in the current year than the previous year. While this is acrude dichotomous measure, such an approach is not uncommon (e.g., Staw et al., 1983;Beattie and Jones, 1992; Clatworthy and Jones, 2001, 2003; Smith and Taffler, 1992a, b,1995).A single year (2000) of data was examined for two reasons: (1) to eliminate the potentialconfounding effects of changes in reporting rules over time; and (2) to avoid the post-Enron period when significant changes in disclosure practice and behaviour were takingplace. Only press releases published before the end of July 2001 were included in theresearch, three months before the financial scandals started (Enron was exposed inOctober 2001) and five months before the first cautionary advice was issued by the SECin December 2001.Measuring impression managementFour content analysis approaches are adopted in this study: (1) A thematic, form-orientedanalysis based on keywords, statements and amounts; (2) Analysis of selectivity ofquantitative information; (3) Analysis of three visual/presentation techniques toemphasise including (a) the location, positioning and visual presentation of disclosures;(b) emphasis by repetition; and (c) emphasis by reinforcing disclosures and (4) Use ofperformance comparisons. These four techniques lend themselves to manual contentanalysis of disclosures and as such form a methodologically cognate cluster. The analysisdistinguishes between quantitative disclosures in accounting narratives and qualitative 13
  • 16. disclosures. Figure 1, which is based on a similar figure in Beattie et al. (2004),summarises the methods of measuring impression management adopted in this study andspecifies whether the technique is applied to quantitative or to qualitative disclosures. Thecoding categories for each impression management method is also shown.Manual content analysisDuriau et al. (2007) performed a content analysis on the content analysis literature inorganisation studies. Having listed the advantages of computer aided textual analysis overmanual methods (larger data sets, reliability, speed, lower cost), they express surprise thatonly 24 of the 98 papers they analysed report using computers for part or all of thecontent analysis. Morris (1994) has tested the validity and reliability of manual andcomputerised approaches. She found that computerised and manual results agreed at anacceptable level, and that computerised coding achieved an acceptable level of semanticvalidity. Conversely, in a more recent study, Conway (2007) found that human andcomputer-assisted coding yielded significantly different results in a content analysis ofnewspaper coverage of a political campaign. He observes that several subjective stepshave to be taken to adapt the content to the program, and that those decisions can be 14
  • 17. arbitrary and fall outside the concept of traditional intercoder reliability. Conway (2007,p. 187), referring to the work of Linderman (2001) states: “Linderman concluded thatcomparing human and computer-assisted coding depends on the complexity ofcategories, with computers working best when categories are ‘easy to operationalize’,but human coders working better with complex categories.” It is our contention thatimpression management techniques are subtle and sophisticated, and therefore complex,and warrant manual content analysis.Previous studies dealing with the content of accounting narratives have used computerprogrammes (e.g., Ingram and Frazier, 1983; Frazier et al., 1984; Tennyson et al., 1990;Smith and Taffler, 2000; Rutherford, 2005; Henry, 2008; Matsumoto et al., 2006; Daviset al., 2007) or a mixture of manual and computer coding (e.g., Smith and Taffler, 1992a;Subramanian et al., 1993; Abrahamson and Park, 1994; Abrahamson and Amir, 1996;Smith and Taffler, 2000). Others have done all the coding manually (e.g., Bettman andWeitz, 1983; Staw et al., 1983; Salancik and Meindl, 1984; Courtis, 1986; Jones, 1988;Lang and Lundholm, 2000; Clatworthy and Jones, 2003). Thus, manual and computerisedanalyses are not necessarily alternatives. In many cases both approaches have been usedtogether.For thematic analysis, computer-based techniques typically rely on software to list thefrequency of occurrence of words which are afterwards coded by researchers (for example,Abrahamson and Park, 1994). Computerised analysis generally requires lists of keywordsto be assembled in advance, and is not as adept at classifying keywords depending oncontext. In manual analysis, the researcher codes keywords and statements directly fromthe content of the text analysed. Given the virtually inexhaustible list of possible keywords,manual coding is arguably more reliable than a computerised approach (Wallace, 1992).Judgement is required in applying coding methods in thematic analysis. Even greatersubjective judgement is required to analyse selectivity, visual/presentation effects(emphasis) and performance comparisons.Similar to other corporate documents, the variety of formats of press releases could beproblematic using computerised coding of visual/presentation effects (emphasis). Forexample, some ARPRs include a headline, others do not; some contain only one paragraph, 15
  • 18. others contain many paragraphs. A further complication in measuring selectivity is that thedata comes from two sources: ARPRs and annual reports. Given the limitations ofcomputerised coding in a complex data set, manual analysis is used in the current study.Manual content analysis is labour-intensive and time-consuming, which limits samplesizes. However, content analysis allows more detailed and sophisticated analysis andcomparisons. It has been subject to criticism due to low validity and reliability arising fromthe exercise of subjectivity in manual coding. These criticisms are considered further on inthe paper.Reliability and validity of codingReliability in the context of content analysis refers to the amount of intercoder agreementbetween multiple coders of the same text. Krippendorff (1980) identifies three types ofreliability: (i) stability, the extent to which the analysis remains unchanged over time; (ii)reproducibility, the degree to which the analysis can be recreated using differentindividuals; and finally (iii) accuracy, the degree to which the analysis conforms to astandard. Validity, on the other hand, refers to the appropriateness of the conclusions,given the content analysis methodology adopted. Morris (1994) describes four types ofvalidity: (i) construct validity, the extent to which the content analysis variables arecorrelated with other measures of the same construct; (ii) hypothesis validity, the extentto which the content analysis variables behave as they are supposed to in relation to othervariables; (iii) face validity, the extent to which the method appears to measure theconstruct it is intended to measure; and (iv) semantic validity, the extent to which personsfamiliar with the language and texts agree with the list of words placed in the categoryhave similar meanings or connotations. Thus, the validity of the underlying classificationis dependent on researchers’ knowledge and experience of the domain being investigated.The test for reliability used in this study is consistent with Clatworthy and Jones (2003),who used a pre-sample of 20 as recommended by Krippendorff (1980) and Breton andTaffler (2001). In this study, reliability of the coding process for qualitative data wastested as follows: a pre-sample of 20 press releases was coded by two independentresearchers and the results were compared with recommended reliability levels. The first 16
  • 19. coder is one of the authors; the second coder is a researcher with a background in masscommunications.Of the 20 press releases, ten were selected randomly among good news companies andten among bad news companies. The second coder was provided with coding instructionsprepared by the first coder1,2, copies of the 20 ARPRs, category definitions(positive/negative keywords/statements), and a form for recording the number of items ineach category for each ARPR. Results of the two coders were compared with differencesteased out and instructions/definitions refined to ensure consistent coding of the entiresample.Content analysis techniques involving quantitative disclosures (selectivity, performancecomparisons) were not tested for reliability as these disclosures are considered to becapable of more objective coding.Assessing achieved reliabilityThe coding agreement rate can be calculated using different methods. The simplestmeasure is the coefficient of agreement, the ratio of the number of pairwise intercoderagreements to the total number of pairwise judgements (Milne and Adler, 1999). Thepercentage of agreement between coders recommended in the literature ranges from 90%(Clatworthy and Jones, 2003) to 80% (Hackston and Milne, 1996; Milne and Adler,1999). One of the limitations of this way of calculating reliability is that, as the number ofcoding categories becomes fewer, the likelihood of random agreement increases. Thus,the coefficient of agreement measure will tend to overestimate the coders’ reliability andthis overestimation increases with fewer categories. In this study, as shown in Figure 1,the number of categories for each impression management method is generally relativelylow (generally two to three categories) and therefore the likelihood of random agreementis high.In this study, the coefficient of agreement is calculated following Milne and Adler(1999). The correspondence between the two coders was high. Overall, the first coderidentified 1,729 items in the impression management methods and categories(keywords/statements, emphasis by location/positioning/visual presentation, emphasis by 17
  • 20. repetition, emphasis by reinforcement) in the sample of 20 ARPRs analysed, while thesecond coder identified 1,681 items, resulting in a concordance of over 95%.However, agreement for two of the impression management methods was low. The areaof most disagreement concerned emphasis by repetition. For example, whereas the firstcoder identified 29 statements (27 positive and two negative) repeated in the sample of 20ARPRs, the second coder found only six (five positive and one negative) resulting in anagreement of only 21%. This may be due to: (1) small counts in some categories (i.e.,number of repeated negative statements coded by the first coder is two and by the secondcoder, one; this results in marginal disagreement of 50%); (2) coding rules are notsufficiently clear. This issue was discussed between the two coders and all cases wereidentified, studied and analysed separately. The coding rules were revised and rewrittenas recommended by Weber (1990, p. 23).Another area of lower than acceptable agreement arose in coding emphasis bylocation/positioning/visual presentation of disclosures. Coding of thelocation/positioning/visual presentation of keywords (64% agreement) and statements(58% agreement) did not reach the rate of agreement recommended in prior literature.This is due to variability in coders identifying the length of the sections in press releases.For example, if one of the coders identifies a longer/shorter most-emphasised section, thenumber of keywords/statements included in this section may vary greatly. For example,in the case of Ashtenne Holdings PLC ARPR 2000, one of the coders identified the most-emphasised section of the press release to be one single statement (four words length),whereas the other coder considered the most-emphasised section as being longer,including a set of five bullet points (114 words length). In this case, the approach taken toresolve disagreement was the same as for repetition of statements (i.e., discussion,analysis, and refinement), and the coding rules for location/positioning/visualpresentation were re-written.In order to resolve disagreement encountered in two of the impression managementmethods/categories coded (emphasis by repetition and emphasis bylocation/positioning/visual presentation) the following actions were taken: (1) the codingrules were revised and changed to promote greater consistency following suggestions 18
  • 21. from the second coder, (2) all cases of disagreement were checked and resolved throughdiscussion and (3) using the revised coding rules, a different random sample of 20 pressreleases was selected and the coding of the two impression managementmethods/categories with high disagreement (repetition and visual emphasis) was repeatedby both coders. Results from this second coding were satisfactory. The rate of agreementachieved for emphasis by repetition of statements was 95%, while the rate of agreementfor emphasis by location/positioning/visual presentation reached 84%.Thematic analysis in press releasesThematic analysis involves analysing texts for themes or tones of expression. In contentanalysis, the unit of analysis can be a word, sentence, theme, paragraph or even the wholetext (Weber, 1990, p. 22). Individual words have no meaning without a sentence orsentences for context. Hence, there are reservations about computerised keywordsearches. Although computers speed up the coding of reports, they may also includecoding mistakes. In relation to the unit of analysis, Milne and Adler (1999) distinguishbetween units used as a basis of coding, versus units of analysis used for measuringdisclosure. They observe that as a basis for coding, sentences have been shown to bemore reliable than any other unit of analysis. Most social and environmental contentanalyses use sentences as their unit of analysis. The same unit of analysis is usually notused for both coding and measuring, with measuring being more commonly based onwords.In this study, both keywords and statements are analysed, for two reasons: (1) thisapproach permits a form of methodological triangulation, cross-checking the analysis ofpositive and negative information; and (2) it identifies results where keywords andstatements provide different outcomes. For instance, a statement might include multiplepositive keywords, yet it may be counted as one positive statement. Results from thestudy of both keywords and statements are expected to be similar (as keywords are likelyto influence categorisation of statements) and therefore provide a crosscheck onreliability. 19
  • 22. Analysis of keywordsA keyword is one which implies an outcome for the firm. Following prior literature,positive and negative keywords are identified and coded. A word was coded asnegative/positive under two conditions (Abrahamson and Park, 1994; Abrahamson andAmir, 1996): (1) The sentence in which it is mentioned includes a negative/positiveoutcome for the company; or (2) The sentence mentions the environment affecting thecompany negatively/positively.As shown in Table 1, thematic analysis has been carried out by reference to keywords(Tennyson et al., 1990; Abrahamson and Park, 1994; Abrahamson and Amir, 1996; Smithand Taffler, 2000; Clatworthy and Jones, 2003; Rutherford, 2005; Henry, 2008),tone/language (Lang and Lundholm, 2000; Davis et al., 2006), and using other constructs(Smith and Taffler, 1995; Lang and Lundholm, 2000; Clatworthy and Jones, 2006;Henry, 2008). The methodology of Tennyson et al. (1990) is complex and includesanalysis of the relationships of the frequencies of words, and factor analysis ofinterrelated words into themes. Starting with Weber’s initial content analysis dictionary,Smith and Taffler (2000) used the Oxford Concordance Program to augment and generatea dictionary of 168 words. They also analysed linear additive composite variables. Theirmeasures are scaled by the number of words in the narrative. Sentences were alsoanalysed thematically.This research uses simpler methods of thematic analysis. A list of keywords wasdeveloped, starting with lists from prior research. Three such lists were used: Clatworthyand Jones (2003) for positive keywords; and Abrahamson and Park (1994), Abrahamsonand Amir (1996) and Clatworthy and Jones (2003) for negative keywords. These listswere added to during the subsequent coding of the sample of 101 press releases,culminating in a final list of 301 keywords. Rutherford (2005) and Henry (2008: 387)also used word lists (90 and 190 words, respectively). All except 18 words3 of the 190words in Henry’s list are common to the lists in this paper. Rutherford’s word list is notcomparable to the list in this paper as he does not classify words between positive andnegative. Unlike other researchers, Rutherford (2005) did not treat grammatical variationsseparately. He combined 16 words that were closely related (i.e., singular and pluralmanifestations). Conversely, Henry (2008) treated similar words separately. 20
  • 23. Table 2 analyses the number of keywords in this study between positive and negative andby reference to their use in prior literature. Of the 301 keywords, 127 (>40%) are uniqueto this research. This is due to a number of factors: (1) the present study includes allkeywords appearing in the press releases, even though the usage frequency might be verylimited (e.g., the words “disruption”, “boosted” and “rockets” appear only once in thesample), (2) similar to Abrahamson and Park (1994), Abrahamson and Amir (1996) andClatworthy and Jones (2003), grammatical variations of words are also counted asseparate keywords (e.g., “extended”, “extending”, “extensions” and “extensive”). Forexample, out of 109 positive keywords new to the current study, 39 are grammaticalvariations of other keywords (i.e., “lead” is a keyword new to this study; however, therewere three grammatical variations also used in the analysis “leader”, “leadership” and“leading”).Table 2: Keywords used in this research Positive Negative TotalKeywords used by both A&P/A&A and C&J None 57 57Keywords used by A&P/A&A only None 2 2Keywords used in C&J only 108 7 115Keywords unique to this study 109 18 127Total number of keywords used in this research 217 84 301A&P/A&A–Abrahamson and Park (1994) / Abrahamson and Amir (1996) (only negative keywords)C&J – Clatworthy and Jones (2003)Following Clatworthy and Jones (2003), coding of a keyword between negative andpositive depends on the context in which it is used. Coding has to be done in context inorder to differentiate between different meanings and connotations of keywords. Toillustrate, Example 1 shows a sentence with three keywords. The words ‘up’ and‘increase’ are both individually considered positive (because they are associated with thephrases ‘profit before tax’ and ‘dividend per share’) while ‘down’ is a negative keyword(because it is associated with the phrase ‘investment profit’). 21
  • 24. Example 1: Keywords (Brixton Estate plc ARPR 2000) Profit before tax up Keyword+1 7.4% to £43.5m; investment profit down Keyword–1 2.2% to £39.6m; total dividend 10.3p per share, an increase Keyword+2 of 3.0%.In Example 2 ‘ahead’ in the first extract means greater than global market growth and iscounted as a keyword. In the second extract, ‘ahead’ is a reference to the following yearand is not counted as a keyword. A simple word count, whether conducted by computeror by person, will not highlight such contextual differences. Example 2: Keywords in context (Aegis Group plc ARPR 2000) Turnover £5,712.5 million, up Keyword+1 19.2% (1999: £4,791.8 million) - ahead Keyword+2 of 8% global market growth… ….which leads me to be optimistic Keyword+3 about our prospects for the year aheadAnalysis of statementsIn this study, a statement is defined as “a cluster of words with different meanings orconnotations that, taken together, refer to some theme or issue” (Weber, 1990, p. 37).This allows for the occurrence of more than one statement within a sentence. Similar tokeywords, statements are classified into positive and negative.The definition of a statement can be ambiguous. Sentence, phrase, statement and themeall have similar meanings and can be used by researchers interchangeably. For example,Clatworthy and Jones (2003) use the terms “sentence” and “statement” interchangeably(confirmed by one of the authors in personal communication). Further, the definition canbe adapted for a particular type of study. For example, Salancik and Meindl (1984, p.245) define causal statements as “those that relate two events by a causal connective orconnective phrase such as ‘caused’, ‘if then’, ‘because’, ‘attributable to’; and so on”.Similarly, Aerts (2001, p. 13) defines statements as “a phrase or a sentence in which acorporate event or performance outcome was linked with a reason or cause for the eventor outcome”. Although, as seen above, “statement” and “sentence” can beinterchangeable, a sentence may include more than one statement. This implies that,where a sentence deals with more than one issue that could be analysed separately, theissues should be treated as separate statements. Example 3 illustrates the inclusion of 22
  • 25. more than one statement within a single sentence. Four statements can be identified fromthe sentence (three positive, one negative). Example 3: Analysis of press releases by statements (Dawson International PLC ARPR 2000) Many challenges lie aheadStatement–1 but we believe we have the foundations to build on the many achievements of the last 12 months Statement+1, deliver our stated goals Statement+2 and remain focused on further improving shareholders valueStatement+3.The first sentence in Example 4 includes three positive keywords (“improved”, “biggest”,“improvement”) and two positive statements. The second sentence includes one positivekeyword (“up”) and one positive statement. The analysis using statements as the unit ofanalysis is different in that there are only three positive statements in total, compared withfour positive keywords. Example 4: Statement including varying numbers of keywords (British Airways PLC ARPR 2000) Passenger yields per RPK improvedKeyword+1 by 7.7 per cent Statement+1, the biggestKeyword+2 year- on-year improvementKeyword+3 since privatisation in 1987 Statement+2. Group turnover for the full year was upKeyword +4 3.8 per cent at £9,278m (£8,940m) Statement+3Table 3 shows a list of words with positive connotations (positive keywords) andnegative connotations (negative keywords) used by TDG Plc. (The actual wording of thepress release is reproduced in Example 5). As previously mentioned, context is importantin the coding process. The word “reduction” is a negative keyword because it isassociated with the word “profits”. Although the company experienced a decrease inprofit (Profit after tax fell from £15.7 million in 1999 to £9.2 million in 2000), the list ofkeywords in Table 3 shows how the company creates a positive impression. There areonly three negative keywords whereas there are 15 positive keywords (one of which,“up”, is mentioned five times) included in the press release. 23
  • 26. Table 3: List of keywords TDG Plc ARPR 2000 Positive Negative 1. Growth Keyword+1 1. Reduction Keyword–1 2. Up (x 5 times) Keyword+2,3,5,6,9 2. Decline Keyword–2 3. Excellent Keyword+4 3. Down Keyword–3 4. Improved Keyword+7 5. Increased Keyword+8 6. Significance Keyword+10 7. Grow Keyword+11 8. Profitably Keyword+12 9. Effective Keyword+13 10. Strong Keyword+14 11. Progress Keyword+15 Key: Keyword + : Positive; –: NegativeExample 5 also shows the analysis of the press release into positive and negativestatements. The analysis of statements shows similar results to that using keywords. Thenumber of positive statements is much higher (at 12) compared with the number ofnegative statements (at two). Although profits fell, the higher number of positive thannegative statements implies a more positive performance than the actual underlyingfinancial position of the company would suggest.Analysis of amountsThematic analysis is also applied to quantitative amounts included in narrativedisclosures in ARPRs. An amount is coded as positive or negative depending on whetherthe current year amount is higher or lower than the prior year amount. Categorisation ofamounts into positive or negative is only possible where the comparative amount or anexplicit statement of the direction of the item is provided in the ARPR itself.Analysis of selectivity in press releasesCompanies are expected to be selective in the financial amounts they disclose in pressreleases, choosing higher profit/earnings per share numbers from the range of numbersavailable for disclosure because this shows a better picture. Wolseley plc illustrates thesecond type of selectivity in Example 6 – picking the “better” profit amount from aselection of profit numbers which could be used in the press release. 24
  • 27. Example 5: Identification of positive and negative keywords and statements in press release(TDG Plc ARPR 2000)STRONG GROWTHKeyword+1 IN CONTRACT LOGISTICSStatement+1 • Group turnover upKeyword+2 7% to £456 million (upKeyword+3 9% at constant exchange rates)Statement +2 • ExcellentKeyword+4 performance from core Contract Logistics business across EuropeStatement+3 - Turnover upKeyword+5 14%Statement+4 and Operating profit upKeyword+6 18%Statement+5 • Profit reductionKeyword–1 in Storage & Distribution due to first half declineKeyword–2 in cold store utilisation.Statement–1 Second half performance significantly improvedKeyword+7 Statement+6 • Overall, headline profit before tax downKeyword–3 6% to £24.7 millionStatement–2 • Final dividend increasedKeyword+8 by 5% to 7.4p, giving 12.4p for the year, upKeyword+9 5%Statement+7 • Announcement today of alliance with Eagle Global LogisticsNeutral statementDavid Garman, Chief Executive of TDG, commented:‘In December 1999, we announced a forward strategy to transform TDG from a predominantly UKand asset-based business, to a truly Europe-wide, solutions based provider of logistics services. Ourresults for the year reflect the significanceKeyword+10 and scale of our transformation during thistransition period.Statement+8We have demonstrated that we can growKeyword+11 our Contract Logistics businesses rapidly andprofitablyKeyword+12 Statement+9. We have also taken effectiveKeyword+13 action to resolve issues in parts ofour Storage & Distribution businessStatement+10. We have a strongKeyword+14 management teamStatement+11who are focused on delivery, and I am looking forward to reporting on further progressKeyword+15 in2001. Statement+12Note: Positive statements 3 and 9 are considered repetitions of positive statement 1. Repetition isdiscussed later in the paper. Example 6: Selection of profit and EPS figures for inclusion in the ARPR from the P&L account (Wolseley plc ARPR 2000) Profit figures Group trading profit up £60 million (19.2%) to £373.2 million EPS figures Earnings per share before exceptionals and goodwill amortisation up 11.0% to 42.26 pence Note: See full profit and loss account of Wolseley in Appendix 1 which shows the full choice of profit figures and EPS figures from which the above selections were madeWolseley plc performed relatively poorly in 2000 (Profit after tax and minority interests1999: £198.9 million; 2000: £193.5 million). Wolseley plc’s full profit and loss account isshown in Appendix 1. It contains ten profit numbers (marked from to in Appendix1) and three earnings per share amounts (marked from to in Appendix 1) on the faceof the profit and loss account from which to select for disclosure in the press release.Influences on the number to select are likely to be two fold: the absolute amount of profitand the increase shown by that amount over the prior year amount. Of the ten profitnumbers, four (from profit on ordinary activities before tax onwards to ) show adecrease over the previous year. The profit figure selected by the company for inclusion 25
  • 28. in the press release (Operating profit) is the second largest amount (in absolute terms) ofthe ten profit figures in the profit and loss account showing an increase of 19.2% over theprior year. Had the largest absolute amount been chosen, this would only have shown anincrease of 16.8% over the prior year. This possibly accounts for the choice of secondhighest profit amount. The only amount that could have shown a better picture wouldhave been profit before goodwill amortisation and after exceptional items , which showsan increase of 21.1% over the previous year. Profit before tax and interest is notselected, possibly as this only shows an increase over the previous year of 6.6%.The earnings per share figure selected for inclusion in the press release is the largest ofthe three earnings per share figures on the face of the profit and loss account, and is theonly one that shows an increase over the prior year amount. The profit amount selected isafter goodwill, exceptionals and before loss on disposal of discontinued operations.However, the earnings per share amount selected is before goodwill and exceptionals.To convert selectivity to a common-size measure, regardless of the number ofprofit/earnings per share amounts disclosed in the profit and loss account, the followingapproach is adopted in this study. All profit and EPS figures reported on the face of theprofit and loss account are ranked from the lowest amount to the highest amount, basedon monetary value (see illustration in Appendix 1). The amount selected for inclusion inthe press release is identified. The amount chosen for inclusion in the press release isassigned to one of three categories of selectivity, High, Medium, Low. Figure 2 illustratesthe categorisation of selectivity, assuming ten earnings amounts are disclosed in the profitand loss account (i.e., the number of profit amounts in Wolseley plc’s profit and lossaccount in Appendix 1). 26
  • 29. Figure 2: Measuring selectivity: Assigning categories No. amounts Ranking 1 2 3 } High 4 5 6 7 } Medium 8 9 10 } Low Example 7: Selection of performance figures from the P&L account (Barclays PLC ARPR 2000) • Operating profit rose 21% to £3,580 million from £2,964 million • Exceptional items of £214 million, up from a deficit in 1999 of £138 million. This includes sale of Dial and Barclays Property Investment Management • Profit before tax up 42% to £3,496 million from £2,455 million • Business as usual costs savings of £260 million • Woolwich acquisition expected to lead to pre tax synergies of more than £400 million per annum by 2004, up from forecast £240 million • Earnings per share based on operating profit above, up to 163.6p from 143.6p • Dividend per share up 16% to 58.0p from 50.0p • £26.3 million donated to the community Operating profit shown above includes the results of The Woolwich from 25th October 2000. It excludes the 1999 and 2000 restructuring charges, goodwill amortisation and costs directly associated with the integration of The Woolwich. Earnings per share based on this operating profit also exclude exceptional items.In addition to selecting from the financial statements the best numbers to disclose,companies may disclose amounts that are not reported in the financial statements(although they may appear elsewhere in the annual report). The following exampleillustrates this practice. In Example 7, Barclays’ ARPR includes operating profit of£3,580 million for the year 2000. This information is shown as the first item, in bulletpoint, at the beginning of the press release. Further down in the press release (in the least-emphasised section – see below for a discussion of least/most-emphasised), the operatingprofit of £3,580 million is shown to include the results of an acquisition and to excludethe 1999 and 2000 restructuring charges, goodwill amortisation and costs directlyassociated with the integration of this acquisition. The actual operating profit on the faceof the profit and loss account is £3,290 million, £290 million lower than the number 27
  • 30. disclosed in the press release. The operating profit included in the ARPR is what wouldbe referred to as a pro forma earnings number in the US literature.Analysis of emphasis in press releasesEmphasis as an impression management tool assumes that the reader notices theinformation emphasised more. Emphasis is analysed in three different ways in this study.Firstly the location/positioning of the disclosures are analysed, which analysis is alsoinfluenced by the visual presentation techniques used in displaying the disclosures (visualemphasis or degree of prominence). Secondly, the use of repetition to emphasise anumber is analysed. Finally, the use of reinforcement to emphasise qualitative disclosuresis analysed. This is the first time this impression management technique has been studied.Empirical studies in accounting examined/investigated the importance of the location orthe order in which information appears in company reports. Staw et al. (1983) investigatethe location of positive and negative information in accounting narratives. Bowen et al.(2005) investigate the extent to which managers place performance metrics strategicallywithin their earnings press releases. They compare the placement of two metrics: GAAPvs. pro forma information. Their results confirm that managers emphasise the metric thatportrays better firm performance by giving that metric a more prominent location in thepress release. Thus, the figure showing better performance (GAAP or non-GAAP figure)is more likely to appear in an earlier section (headline or first paragraph) of the pressrelease, whereas the figure showing worse performance is buried down in the main bodyof the press release.To analyse visual emphasis, sections of each press release are assigned three levels ofemphasis: (1) most-emphasised, (2) next-most-emphasised and (3) least-emphasised (seeFigure 3). This methodology is adapted from (Staw et al., 1983; Bowen et al., 2005).Visual emphasis is defined as the emphasis provided by prominent location/positioning(e.g., heading and subheadings), special character (e.g., bullet points), type of font (e.g.,bold, italics, underlining, colour), or a combination of two or more of these (Figure 3illustrates these varying degrees of prominence). Press releases do not have a standardpresentation/format. For this reason, classification of disclosures between the threecategories (most-emphasised, next-most-emphasised and least-emphasised) requires 28
  • 31. judgement. Some press releases use explicit visual emphasis techniques while others donot. Where no visual emphasis is used, three levels of emphasis (most-emphasised, next-most-emphasised and least-emphasised) are defined by reference to thelocation/positioning of information in the press release following Bowen et al. (2005).The basic idea is that earlier text in press releases is given greater emphasis simplybecause it comes first. Paragraphs one and two are considered to be the most-emphasised,paragraphs three and four are the next-most-emphasised, and information after paragraphfour is considered the least-emphasised. Press releases which use these emphasistechniques may include three degrees of visual emphasis: (a) a press release with onlyone form of emphasis (for example, headline) followed by plain text. In this case, thesection of the press release with the visual emphasis is coded as the most-emphasisedsection while the methodology used by Bowen et al. (2005) is applied to the plain text.Therefore, the first and second paragraphs are coded as next-most-emphasised section,and the remainder of the text is coded as the least-emphasised section; (b) the textpresents at least four (for example, headline, subheadings, bullet points and bold text) ofthe methods described in Figure 3. In this case, the headline is considered the most-emphasised section of the press release. Use of subheadings, bullet points or bold text iscoded as next-most-emphasised section, and the plain text is coded as least-emphasisedsection of the press release; and (c) when either two or three methods of emphasis fromthose described in Figure 3 are used. In this situation, the highest of the four methodsidentified in Figure 3 is coded as most-emphasised section. Plain text is coded as theleast-emphasised section of the press release and anything in between is the next-most-emphasised section of the press release. Figure 3: Visual emphasis or degree of prominence Ranking of emphasis Most emphasised Headline Subheadings Bullet points Bold text Plain text Least emphasised 29
  • 32. Emphasis – Location/positioning/presentation of qualitative informationVisual emphasis refers to the location or positioning of disclosures in the press release.Location/positioning may depend on whether the disclosure refers to positive or negativeevents or outcomes. For example, a company might want to include negative informationin a less prominent location. Example 8 illustrates the approach taken in coding a pressrelease into the three location/positioning categories (most-, next-most, least-emphasised). As shown in Example 8, even though TDG Plc is a bad news company,Positive Statement 1 (a positive statement by inclusion of a positive keyword) is theheadline in the press release. Although two negative statements are included in the next-most-emphasised section of the press release, the positive statements are exaggerated bytheir inclusion in the headline. The two negative statements are under-stated by beinglocated further down in the press release.Example 8: Different emphasis by location/positioning/visual presentation of positive/negativestatements (TDG Plc ARPR 2000)Text Location of textSTRONG GROWTHKeyword+1 IN CONTRACT LOGISTICSStatement+1 Most-emphasised • Group turnover upKeyword+2 7% to £456 million (upKeyword+3 9% at constant exchange rates)Statement+2 • ExcellentKeyword+4 performance from core Contract Logistics business across EuropeStatement+3 – Turnover upKeyword+5 14%Statement+4 and Operating profit upKeyword+6 18%Statement+5 • Profit reductionKeyword–1 in Storage & Distribution due to first half Next-most- declineKeyword–2 in cold store utilisation.Statement–1 Second half performance emphasised significantly improvedKeyword+7 Statement+6 • Overall, headline profit before tax downKeyword–3 6% to £24.7 millionStatement–2 • Final dividend increasedKeyword+8 by 5% to 7.4p, giving 12.4p for the year, upKeyword+9 5%Statement+7 • Announcement today of alliance with Eagle Global LogisticsNeutral statementDavid Garman, Chief Executive of TDG, commented:‘In December 1999, we announced a forward strategy to transform TDG from apredominantly UK and asset-based business, to a truly Europe-wide, solutionsbased provider of logistics services. Our results for the year reflect thesignificanceKeyword+10 and scale of our transformation during this transitionperiod.Statement+8 Least-emphasisedWe have demonstrated that we can growKeyword+11 our Contract Logisticsbusinesses rapidly and profitablyKeyword+12 Statement+9. We have also takeneffectiveKeyword+13 action to resolve issues in parts of our Storage & Distributionbusiness.Statement+10 We have a strongKeyword+14 management teamStatement+11 who arefocused on delivery, and I am looking forward to reporting on furtherprogressKeyword+15 in 2001.Statement+12 30
  • 33. Emphasis – Location/positioning/presentation of quantitative informationMost ARPRs studied disclose positive amounts in the most-emphasised section of thepress release. Few of the press releases include one or more negative amounts in themost-emphasised section. In Example 9, QXL ricardo plc includes seven quantitativeitems. Although the company performed poorly (Losses after tax before minorities2000/01: £(143.1) million; 1999/00: £(66.7) million), six of the seven disclosures arepositive. Only one negative item (Quantitative item 7) is disclosed in the press release.This item is placed in the least-emphasised section after disclosure of all positivequantitative items. Example 9: Quantitative information and visual emphasis (QXL ricardo plc ARPR 2000) Most-emphasised section No quantitative items included Next-most-emphasised section Quantitative item 1: Positive quantitative item Growth in agency-based Gross Auction Value of 42% Quantitative item 2: Positive quantitative item Gross profit increased 35% Least-emphasised section Quantitative item 3: Positive quantitative item A 319% increase in Gross Auction Value to £89.3 million for the year, compared to £21.3 million for the year ended 31 March 2000 Quantitative item 4: Positive quantitative item Total members increased to 2.9 million at 31 March 2001, a 415% increase compared to 557,000 at 31 March 2000 Quantitative item 5: Positive quantitative item Number of items listed for auction increased to 31.7 million for the year ended 31 March 2001, a 484% increase compared to 5.4 million for the year ended 31 March 2000 Quantitative item 6: Positive quantitative item Gross profit up to £2.6 million, an increase of 258% over £741,000 for the year ended 31 March 2000 Quantitative item 7: Negative quantitative item Trading loss of £49.4 million, compared to a loss of £32.8 million for the year ended 31 March 2000.Emphasis – Repetition of qualitative informationRepetition of information can enhance the understandability of financial reports or it canadd noise to the reporting process (Courtis, 1996). Courtis (1996) investigates the 31
  • 34. presence of superfluous disclosures in Hong Kong annual reports by testing redundantdisclosure (such as repetition) against some corporate attributes (size, profitability, riskand industrial grouping). For the purposes of this study, repetition is said to occur when apress release includes the same piece of information more than once. A statement isdeemed to be repeated even where there is slight variation in one or two words in the twostatements. This technique can be misleading for two reasons: (1) the press release is ashort document (2 pages on average) and repetition of the same issue can cause the readerto focus on that specific issue while diverting attention from other issues in the pressrelease and (2) this practice can be misleading if the manager repeats positive informationbut not negative or vice versa. An example of repetition of a statement has already beenpresented in Example 5 where a single piece of information (positive information) wasrepeated three times. Example 10 also illustrates this practice where substantial newbusiness is emphasised by repetition. The reference to new business in the headline isexacerbated by two repetitions of this positive information. This suggests that it is notenough to look at individual impression management techniques in isolation. Theinteraction effects of using two or more impression management techniques at the onetime must be considered. Example 10: Multiple repetition of positive statements (Aegis Group plc ARPR 2000) Headline Record new business performance reflects effective strategy First repetition in main body Record new media business wins totalling $2,050 million (1999:$1,206 million) Second repetition in main body 2000 was a good year for Aegis with a record $2 billion of new media business won during the yearEmphasis – Repetition of quantitative informationAs is the case with qualitative information, repetition of quantitative information is alsocommon practice. This consists of reiterating the same amount in the same press release.The three statements shown in Example 11 are included in different parts of the pressrelease. The first one is in the headline. The second is the first repetition occurring in themain body of the press release. The third statement is the second repetition of the samequantitative item. 32
  • 35. Example 11: Repetition of quantitative items (British Airways PLC ARPR 2000) Headline Full year profit of £150 million First repetition in main body Full year pre-tax profit of £150 million, up from £5 million a year ago Second repetition in main body British Airways today posted a pre-tax profit of £150 million for the 12 months ended March 31, 2001 (2000: £5 million)Emphasis – Reinforcement of qualitative informationReinforcement occurs when emphasis is added to a particular keyword by use of aqualifier. Although it is not a pervasive practice, its inclusion in this study is consideredimportant as evidence of one of the disclosure practices used by managers whenpreparing their reports. This impression management technique has not been studied inprior literature. TDG Plc ARPR 2000 (Example 12), a bad news company (profit incurrent year lower than prior year), discloses four reinforcements. One of them(Reinforcement 3) is a double reinforcement where two qualifiers “rapidly” and“profitably” reinforce one positive keyword “grow”. The other three are reinforcementsof positive keywords. The words “strong” (Reinforcement 1), “significantly”(Reinforcement 2) and “further” (Reinforcement 4), reinforce the positive keywords“growth”, “improved” and “progress”, respectively. Example 12: Multiple reinforcement of positive keywords (TDG Plc ARPR 2000) Reinforcement 1: Reinforcement of positive keyword (headline) Strong growth in contract logistics Reinforcement 2: Reinforcement of positive keyword (main body) Second half performance significantly improved Reinforcement 3: Double reinforcement of positive keyword (main body) We have demonstrated that we can grow our Contract Logistics businesses rapidly and profitably Reinforcement 4: Reinforcement of positive keyword (main body) I am looking forward to reporting on further progress in 2001Another method falling within the technique of reinforcement of qualitative informationis diminution of keywords. This occurs when a keyword is accompanied by a qualifierwhich lightens its effect. This practice is more likely to occur with negative keywordsthan with positive keywords. Example 13 illustrates this issue. The qualifier “a little” de-emphasises the negative connotation of the keyword “fallen”. 33
  • 36. Example 13: Diminution of negative keywords (Silentnight Holdings Plc ARPR 2000) Since then demand has fallen a little, particularly for cabinet and upholstered furnitureIn Example 14, the word “slightly” de-emphasises the word “ahead”, showing that thecompany met its forecast, without exceeding it too much which might suggest inaccurateforecasting by the company. Example 14: Diminution of positive keywords (Uniq plc ARPR 2000) Profit before tax, exceptional items and goodwill amortisation of £57.5m, slightly ahead of profit forecast made in March 2001Analysis of performance comparisons in press releasesPrior research has studied performance comparisons from the point of view of managersselecting performance comparisons that allow the best performance to be portrayed. Welook at performance from a different perspective - as a means of reinforcing quantitativeinformation in press releases. Quantitative information about performance can beprovided in terms of monetary and non-monetary amounts. Non-monetary quantities arealso coded, depending on their format/presentation, e.g., numbers, percentages.Percentages are used with benchmarks of company performance related to either priorperiod(s) or industry performance. Such benchmarks are commonly provided in pressreleases. The percentage might be disclosed on its own or together with a current yearmonetary amount.Quantified monetary amounts with comparisons are classified into three categories –positive, negative and neutral amounts – depending on whether the amount highlightedhas increased, decreased or remained unchanged by reference to the comparator.Performance comparisons are a form of emphasis in that they reinforce a quantitativeamount. Reinforcement of quantitative items occurs when managers disclose: (1) abenchmark indicating the percentage change over the prior year together with the currentyear amount; or (2) the amount from the prior year together with the current year figure.The company can also choose to report both: a benchmark in the form of a percentageand the amount from the prior year together with the current year figure. The latter 34
  • 37. represents double emphasis, whereas options (1) and (2) are single emphasis. Example 15illustrates these practices. Example 15: Reinforcement of quantitative items: single emphasis (Wolseley plc ARPR 2000) Reinforcement including a benchmark in the form of percentage Pre-tax profit before exceptionals and goodwill amortisation up 9.4% to £357.4 million Reinforcement including prior year amount £288 million (1999: £310 million) invested in acquisitionsDouble emphasis allows the reader to cross-check the calculation and is more transparent,but less common. Categorisation into positive/negative amount is only possible where thecomparative amount is provided in the ARPR itself (Hoskin et al., 1986). Wherecategorisation into positive/negative is not possible, the amount is deemed to be neutral.In general, we expect management to disclose performance comparators that reflect thecompany’s performance in a positive rather than negative light.For example, TDG Plc includes six monetary and non-monetary disclosures in its pressrelease, five of which are positive figures and only one negative. Quantitative item 1 inExample 16 is positive (increase in Group turnover). Quantitative items 2 and 3 show anincrease in business segment turnover and operating profit. One negative quantitativeitem (Quantitative item 4) shows a decrease in headline profit before tax (beforeexceptional items and amortisation). In Quantitative items 5 and 6, final dividend anddividend for the year are shown to have increased by 5%. Example 16: Reinforcement of quantitative items included in the ARPR (TDG Plc ARPR 2000) Quantitative item 1: Reinforcement of positive quantitative item Group turnover up 7% to £456 million Quantitative items 2 and 3: Reinforcement of positive quantitative item (two quantitative items) Excellent performance from core Contract Logistics business across Europe - turnover up 14% and operating profit up 18% Quantitative item 4: Reinforcement of negative quantitative item Headline profit before tax down 6% to £24.7 million Quantitative items 5 and 6: Reinforcement of positive quantitative item (two quantitative items) Final dividend increased by 5% to 7.7p, giving 12.4p for the year, up 5% 35
  • 38. Use of comparisons is also illustrated in Example 6 earlier. Although Wolseley plcperformed relatively poorly in 2000 (Profit after tax and minority interests 1999: £198.9million; 2000: £193.5 million) the two performance numbers (profit and earnings pershare) selected for disclosure show substantial increases against their prior yearcomparators.Companies may be influenced in their use of reinforcement depending on whether thefigures are positive or negative. For example, a company may disclose negativequantitative items without reinforcement but use reinforcement with positive amounts.British Vita (Profit after tax and minority interests 1999: £48.5 million, 2000: £43.3million) performed relatively poorly in 2000. The annual report shows that the profitbefore tax for the current year was only £80.1 million compared with a prior year profit£84.6 million. In Example 17, British Vita PLC does not include a benchmark with theprofit before tax amount disclosed, which hides or disguises the decrease in profit fromthe prior year. Example 17: No reinforcement of negative quantitative item (British Vita PLC ARPR 2000) Profit before Tax of £80.1mCity North Group plc also performed poorly in 2000 (Profit after tax and minorityinterests 1999: £1,768,000, 2000: £715,000). In Example 18, despite poor performance in2000 compared with 1999, City North Group plc reports only positive amounts and allinclude a benchmark showing increases from prior year. Example 18: Reinforcement of positive quantitative items (City North Group plc ARPR 2000) Quantitative item 1: Reinforcement of positive quantitative item Net assets up 22% to £62,400,000 Quantitative item 2: Reinforcement of positive quantitative item Diluted net asset value per share up 21% to 278p Quantitative item 3: Reinforcement of positive quantitative item Rental income up 17% to £4,080,000 Quantitative item 4: Reinforcement of positive quantitative item Operating profit up 17% to £2,390,000 36
  • 39. Companies may disclose in the ARPR a current year amount that best portraysperformance compared with a prior year benchmark. In Example 19, National Grid plcdiscloses profit before tax and exceptional items of £481.3m in its ARPR. This amountdoes not appear in the related group profit and loss account. The amount disclosed of£481.3m is calculated as the profit before exceptional items of £731.9m less net interestof £250.6m, both of which amounts do appear on the face of the group profit and lossaccount. Selecting this idiosyncratic amount of £481.3 allows National Grid plc toprovide the best possible prior year comparative benchmark of £481.6m (Profit beforeexceptional items of £546.5m less Net interest of £64.9m) against which to compare theamount disclosed. This allows management to suggest in the ARPR that performance is“level” compared with the previous year. No other combination of current year and prioryear profit amounts would portray as good a picture of National Grid plc’s performance. Example 19: Selectivity and benchmarking (National Grid plc ARPR 2001 & Annual Report 2000) ARPR 2001 Allowing for the higher interest expense, we held pre-tax profit before exceptional items and goodwill amortisation level at £481.3 million and increased earnings per share on the same basis by 9 per cent. We also had exceptional profits of over £470 million relating to the reduction in our holding in Energis. Extracts from Annual Report 2000 Group profit and loss account for the years ending 31 March 2001 2000 1999 £m £m £m Operating profit – Before exceptional integration costs and goodwill amortisation 731.9 546.5 579.9 – Exceptional integration costs (45.3) – – – Goodwill amortisation (74.5) (7.9) (2.5) Total operating profit – continuing operations 612.1 538.6 577.4 Exceptional profit relating to partial disposal of Energis 242.9 1,027.3 891.8 Profit on disposal of businesses 20.1 – – Net interest (250.6) Exceptional cost of closing out interest rate swaps – – (52.6) Profit on ordinary activities before taxation – continuing operations 624.5 1,501.0 1,298.1 Taxation 149.6 (352.6) (283.1) Profit on ordinary activities after taxation 774.1 1,148.4 1,015.0 Earnings per ordinary share – Basic, including exceptional items and goodwill amortisation 52.1p 78.0p 69.2p – Basic, excluding exceptional items and goodwill amortisation 26.5p 24.3p 23.3p – Diluted, including exceptional items and goodwill amortisation 49.5p 73.4p 65.2p – Diluted, excluding exceptional items and goodwill amortisation 25.8p 23.8p 22.7p 37
  • 40. Comparing disclosure of quantitative and qualitative information in ARPRsCompanies can manipulate quantitative and qualitative information to create impressions.For example, a statement might include a negative quantitative item displayed in neutralterms (without showing the prior year amount). This negative quantitative item can evenbe shown in such a way as to be construed as a positive statement.Skinner (1994) has found that companies are more likely to disclose positive informationin quantitative format and negative information in qualitative format. Consistent with thisfinding, Manganese Bronze Holdings PLC includes four negative statements in its ARPR2000 (Example 20). Each of these four statements is in narrative form, with noquantitative negative items being disclosed. Example 20: Negative information in qualitative format (Manganese Bronze Holdings PLC ARPR 2000) Negative statement 1: Components Division continued to suffer losses Negative statement 2: However, increased production costs have led to reduced margins Negative statement 3: The Components Division again lost money mainly affected by a shortage of sales of sintered components to the motor industry Negative statement 4: If market conditions remain as at present we expect lower profits for our Vehicles Division in the current yearCompanies may also disclose negative items in a neutral way. For example, Stagecoach isclassified as a bad news company (Profit / (loss) after tax and before minority interests1999: £38 million; 2000: (£332) million). Stagecoach Group plc ARPR 2000 (Example21) includes positive information in quantitative and qualitative format and a negativeitem as part of a neutral statement. Example 21: Qualitative and quantitative information (Stagecoach Group plc ARPR 2000/2001) Positive statement and positive quantitative item Turnover excluding discontinued operations £2,067.3 million, up 17.4% Neutral statement and negative quantitative item Profit before tax, goodwill amortisation and exceptional items £122.9 million (2000: £244.3 million) 38
  • 41. Impact of range of impression management techniquesThe 21 examples in this paper, together with Appendix 1, illustrate the myriad oftechniques adopted by companies to present company performance in their narrativereports. It is only possible to obtain a fuller picture of reporting practices by consideringthese techniques together.Section 4 of the paper develops the qualitative and quantitative measures of impressionmanagement into two composite impression management scores using the fourqualitative measures ((i) Thematic – keywords and phrases – number of positive; numberof negative; (ii) Emphasis – Location; (iii) Emphasis – Repetition and (iv) Emphasis –Reinforcement) and the five quantitative measures discussed earlier ((i) Disclosure ofquantitative performance monetary and non-monetary amounts; (ii) Selectivity; (iii)Emphasis – Location; (iv) Emphasis – Repetition and (v) Performance comparisons).4. Constructing a composite impression management scoreComposite scores – unweighted or weighted averages of a number of underlyingvariables – are common in research, e.g., Altman Z-score, disclosure indices, governancescores. The justification for using such scores is that one metric alone can give amisleading picture. Equally, it is difficult to reduce complex corporate processes to asingle measure. Our thesis is that impression management should be measured in aholistic manner, and not merely by using a single measure of impression management.Thus, we have devised two synthesised measures resulting in two composite impressionmanagement scores based on the four qualitative/five quantitative measures referred toabove. Beattie et al. (2004) also develop a holistic measure for analysing narratives inannual reports. They use computer-assisted methods for implementing their four-dimensional framework for holistic content analysis of accounting narratives.Calculation of composite impression management scores for qualitative disclosuresThe qualitative composite impression management score is based on either (i) keywordsor (ii) statements, combined with evidence of three types of emphasis(location/positioning, repetition, reinforcement). Only repetition of statements can bemeasured. Measurement of repetition of keywords is not feasible. A press release caninclude/repeat the same keyword throughout the document without representing 39
  • 42. repetition of the same issue. For example, Millennium & Copthorne Hotels PLC ARPR2000 includes the positive keyword “progress” in the following statements: “Significantprogress in integrating acquisitions” and “Good progress with planned disposal of non-core assets in US”. Although the positive keyword “progress” is repeated, it refers todifferent issues. It would not make sense to consider repetition of this keyword (and otherkeywords) as repetition. In addition, only reinforcement of keywords can be measured.Measuring reinforcement of statements is not practicable. As we define reinforcement asemphasis added to a particular keyword using a qualifier, we cannot apply this to astatement or sentence. The emphasis is put on an individual keyword.Disclosures are made in a hierarchical manner, and the qualitative composite impressionmanagement score includes weightings to capture this hierarchy. Weighting systems arecommon in research but there is no method that is universally accepted. Such weightingsare subjective. It is highly questionable how to weight different elements in a compositeimpression management score. To take account of this subjectivity, it is recommendedthat the weightings be varied in empirical research using a composite impressionmanagement score, and that empirical results be subjected to sensitivity analysis to checkwhether variations in the weightings influence the results. Prior research using weightedand unweighted/naïve disclosure models arrived at similar results (e.g. Chow and Wong-Boren, 1987; Cooke, 1989; Botosan, 1997; Zarzeski, 1996; Riahi-Belkaoui, 1999).Table 4 summarises the weightings to be applied and the resulting calculation ofqualitative composite impression management scores. Weightings in this context aresummative, not multiplicative. Each keyword/statement is given a weighting of 1.0. If thekeyword/statement appears in the most-emphasised section, a weighting of 1.0 is added;for the next-most emphasised section a weighting of 0.5 is added; the least-emphasisedsection attracts no weighting. If the keyword is reinforced a weighting of 0.5 is added. Ifthe statement is repeated, a weighting of 0.5 is added. Gordon et al. (2007) also aggregateindividual elements in calculating a composite score for management credibility and forrestatement announcement characteristics. Similar to the method in this paper, they applya combination of weightings of 1.0 and 0.5 for elements of their aggregated compositescore. 40
  • 43. Applying these weightings, the resulting qualitative composite impression managementscore will vary from a maximum of 2.5 (e.g., where a keyword is included in the most-emphasised section of the ARPR, the statement of which it is part is repeated and thekeyword is reinforced) to a minimum of 1.0 (e.g., where a keyword is included in theleast-emphasised section of the ARPR and is not reinforced, and the statement of which itis part is not repeated). Table 4: Method for calculating qualitative composite impression management scores for keywords/statements Measure Weighting (i) Thematic – keywords/statements 1.0 (ii) Emphasis – Location: Most-, next-most, least-emphasised 1.0/0.5/0.0 (iii) Emphasis – Repetition (Statements only) 0.5 (iv) Emphasis – Reinforcement (Keywords only) 0.5 Maximum possible composite score per keyword/statement 2.5 Minimum possible composite score per keyword/statement 1.0 Note: the score will be either positive (+) or negative (-) depending on whether the keyword is positive/negativeCalculation of composite impression management scores for quantitative disclosuresIn the first instance, quantitative disclosures in ARPRs are given a score of 1.0. The scoreis then adjusted for similar reasons to those applying to qualitative scores.In addition to the effects of location and repetition, the quantitative composite impressionmanagement score varies depending on whether selectivity is measured in respect of thequantitative amount. Similar to reinforcement of statements, measurement ofreinforcement of quantitative amounts is not feasible.Table 5 summarises the weightings to be applied, and the resulting calculation ofquantitative composite impression management scores. Each quantitative amountidentified in the ARPR is given a weighting of 1.0. If the quantitative amount is selectedfrom a range of possible numbers in the profit and loss account, a weighting is addeddepending on the ranking within the profit and loss account of the amount selected. 41
  • 44. If the quantitative amount is selected from the profit and loss account, it is rankeddepending on whether the amount selected is in the highest/medium/lowest category ofamounts from which selection can be made. A weighting of 1.0 is provided for highestcategory; for the medium category a weighting of only 0.5 is added. The lowest categoryattracts no weighting.If the quantitative amount appears in the most-emphasised section of the ARPR, aweighting of one is added; for the next-most emphasised section a weighting of 0.5 isadded; the least-emphasised section attracts no weighting.If the quantitative amount is accompanied by a performance comparison, an additionalweighting of 0.5 is added.If the quantitative amount is repeated, an additional weighting of 0.5 is added.Applying these weightings, the resulting quantitative composite impression managementscore will vary from a maximum of 4.0 to a minimum of 1.0. As with keywords andstatements, “positive” and “negative” numbers disclosed retain their + and – descriptivetags.Using composite impression management scores to measure biasIn addition to measuring impression management, the composite impression managementscores can relatively easily be extended to provide a measure of bias. The impressionmanagement bias score captures the extent impression management is biased towardgood news /optimistic language/ tone. Thus, qualitative and quantitative compositeimpression management scores could be further manipulated to capture a measurementfor bias inherent in impression management, resulting in an impression management biasscore. The impression management bias score is an index and comprises the differencebetween the total composite impression management scores for all positivekeywords/statements/quantitative amounts minus the total composite impressionmanagement score for all negative keywords/statements/quantitative amounts, divided bythe total composite impression management scores for all keywords/statements/ 42
  • 45. quantitative amounts. The impression management bias score can be expressed asfollows:P IM Score - N IM ScoreP IM Score + N IM ScoreWhere P IM Score = total positive impression management score for a press release N IM Score = total negative impression management score for a press releaseTetlock et al. (2008) compute a simple quantitative measure of language (in newspaperarticles) to investigate whether such measures have incremental explanatory power forfirms’ future earnings and stock returns. Their primary measure to quantify the languageused in financial newspapers is the fraction of negative words in a news story. However,they also calculate other variations of the measure based on the differences betweenpositive and negative words divided by the total positive and negative words in thenewspaper article. Henry (2008) calculates Tone as the count of positive words minusnegative words, divided by the sum of positive and negative word counts. Thus, theirmethods are similar to the measure of bias proposed in this study.The impression management bias score /index is illustrated in Table 6. The impressionmanagement measures for quantitative disclosures set out in Table 5 are applied in Table6. For simplicity, only quantitative disclosures are used and no repetition, selectivity orperformance comparison is assumed. The table assumes that seven positive quantitativeamounts and three negative quantitative amounts are disclosed in the press release. Thenegative disclosures are located in the least-emphasised location in the press release,while the positive disclosures appear throughout the press release. The impressionmanagement bias score is measured at +0.57. A score of zero means no bias, and a scoreof +1 mean positive bias only, there being no negative disclosures. Thus, an impressionmanagement bias score of +0.57 suggests strong positive impression management bias(almost 60 per cent of the maximum possible positive bias of +1.0). 43
  • 46. Table 5: Method for calculating quantitative composite impression management scores for amountsMeasure Selectivity applies No selectivity Weighting Weighting(i) Disclosure of quantitative performance monetary and non-monetary amounts 1.0 1.0(ii) Selectivity - highest/medium/lowest category of amounts from which selection can be made 1.0/0.5/0.0(iii) Emphasis – Location: Most-, next-most, least-emphasised 1.0/0.5/0.0 1.0/0.5/0.0(iv) Emphasis – Repetition 0.5 0.5(v) Performance comparisons 0.5 0.5Maximum possible composite score per quantitative amount 4.0 3.0Minimum possible composite score per quantitative amount 1.0 1.0Note: the scores will be either positive (+) or negative (-) depending on whether the amount is positive/negative Table 6: Calculating bias using quantitative disclosuresScenarioAssume the press release on which this example is based has disclosed seven positive quantitative amounts and three negativequantitative amounts. Three positive quantitative amounts are located in the first paragraph (i.e., most-emphasised location) of thepress release. Two positive quantitative amounts are located in the middle of the press release (i.e., next-most-emphasised location),while the remaining two positive quantitative amounts and the three negative quantitative amounts are in the last paragraph (i.e.,least-emphasised location) of the press release. For simplicity, the press release contains no repetition of quantitative amounts, noselectivity and includes no performance comparisons. Measure Positive Negative Total amount amounts amounts Number of quantitative disclosures 7 3 10 Positive Negative Total Composite impression management score score score score (1) Disclosure of quantitative performance monetary and non-monetary amounts 7 3 10 (2)(a) Emphasis – Location: - Most 3x1 0 3 - Next-most 2 x 0.5 0 1 - Least-emphasised 2 x 0.0 3 x 0.0 0 (2)(b) Emphasis – Repetition 0 0 0 (3) Performance comparisons 0 0 0 (4) Selectivity - highest/medium/lowest category of amounts from which selection can be made 0 0 0 Total composite impression management score 11 3 14 Bias score 11Positive composite score –3 Negative composite score = 8Net positive composite score/14 Total composite score = + 0.57 Key: +1 = completely positively biased; –1 = completely negatively biased; 0 = no bias 44
  • 47. 5. Summary and ConclusionsHaving critiqued research on impression management methods in prior financialreporting research, this paper developed and refined the measurement of four less-researched impression management techniques and illustrated these with examples frompress releases. By investigating four impression management techniques in a single paperthe evidence is more complete than in prior studies.Consistent with prior research findings, impression management in financial reportingnarrative disclosures is found, with multiple methods of exercising impressionmanagement in evidence. Company managers have many and varied opportunities toinfluence user impressions. While this research is not systematic, and therefore notgeneralisable to a population, it nonetheless highlights for future researchers theimportance of considering multiple methodologies and more holistic approaches tocontent analysis of narrative disclosures in pursuit of evidence of impressionmanagement.Limitations of the researchThis research is entirely focused on the supply-side, i.e., impression managementtechniques used by management. The demand-side, the users’ perspective, is notconsidered. The key question, does impression management matter, is not addressed. Forexample, we do not know whether users are influenced by these techniques, we do notknow how users’ perceptions are changed by impression management, we do not knowwhether users discount or disregard impression management in corporatecommunications. Some limitations of our research methods have been mentioned earlier.The 21 examples in the paper are drawn from UK press releases for the year 2000.Although these examples are aging, we have no reason to believe they are notrepresentative of corporate practice now as well as in the year 2000.The methodology in this paper recognises that in practice multiple impressionmanagement methods are used simultaneously. Studying four impression managementmethods simultaneously extends and improves on the prior literature. However, themethodology does not include all seven impression management methods. There are 45
  • 48. opportunities in the future for extending the methodology to encompass all sevenimpression management methods.Difficulties in achieving consistency of coding when analysing qualitative data withlinguistic subtleties have been referred to in Section 3. Moreover, the coding rules wereprepared and revised by one of the coders. In addition, although the coding rules were re-written after coding differences between the two coders emerged, the re-written codesshould have been re-tested using additional coders to ensure the guidelines were clear.Suggestions for future researchThe methodology set out in this paper provides researchers with tools to examinemultiple impression management methods simultaneously. A question central toimpression management research is whether the use of impression management variesdepending on firm performance. Or in other words, is there evidence of biased financialreporting, contrary to desirable qualities of good financial reporting as expressed inconceptual frameworks for financial reporting? The development of a impressionmanagement bias score to measure bias arising from the use of impression managementprovides researchers with a method to address this issue.The development of a composite impression management score provides a tool forcomparative research. Composite impression management scores could be computed fornarrative reporting practices across firms, within the same firm over time, across industrysectors, in different disclosure vehicles, and in different jurisdictions.The use of press releases by business journalists is another avenue for enquiry. Dojournalists reproduce narrative disclosures word-for-word from press releases?Comparisons of narrative disclosures in press releases with their reproduction in thefinancial press, using methodologies described in this paper, has rich potential.To conclude, the methods developed offer an extension of measurement techniques toenhance impression management research in the future to address the many un-researched issues thereon. 46
  • 49. Endnotes1 The coding rules comprise the following: (1) Coding rules for second coder; (2)Illustration of qualitative coding of a press release; (3) Illustration of quantitative codingof a press release; (4) List of positive keywords used in the research; (5) List of negativekeywords used in the research. These documents are available from the authors onrequest.2 A weakness of the methodology is that the coding rules were prepared and revised byone of the coders. It would have been better to have stricter delineation (followingKrippendorff, 1980) between the guideline-setter and the coders to ensure reliability ofoutcomes.3 Henry (2008) includes 6 positive words not in the word list in this paper: certain,definitive, delivers, rewards, enjoy, beat. There are 12 negative words not included in theword list: hurdle, obstacle, slump, uncertain, unsettled, risk, threat, penalty, drop, shrink,below, under. 47
  • 50. ReferencesAbrahamson, E. and Amir, E. (1996), "The information content of the presidents letter to shareholders", Journal of Business Finance and Accounting, Vol. 23 No. 8, pp. 1157-1182.Abrahamson, E. and Park, C. (1994), "Concealment of negative organizational outcomes: an Agency Theory perspective", Academy of Management Journal, Vol. 37 No. 5, pp. 1302-1334.Adelberg, A. H. (1979), "Narrative disclosures contained in financial reports: means of communication or manipulation?" Accounting and Business Research, Vol. 9 No. 35, pp. 179-189.Aerts, W. (1994), "On the use of accounting logic as an explanatory category in narrative accounting disclosures", Accounting, Organizations and Society, Vol. 19 No. 4/5, pp. 337-353.Aerts, W. (2001), "Inertia in the attributional content of annual accounting narratives", The European Accounting Review, Vol. 10 No. 1, pp. 3-32.Aerts, W. (2005), "Picking up the pieces: impression management in the retrospective attributional framing of accounting outcomes", Accounting, Organizations and Society, Vol. 30 No. 6, pp. 493-517.Baginski, S. P., Hassell, J. M. and Hillison, W. A. (2000), "Voluntary causal disclosures: Tendencies and capital market reaction", Review of Quantitative Accounting and Finance, Vol. 15 No. 4, pp. 371-389.Baginski, S. P., Hassell, J. M. and Kimbrough, M. D. (2004), "Why do managers explain their earnings forecasts?" Journal of Accounting Research, Vol. 22 No. 1, pp. 1- 29.Baird, J. E. and Zelin, R. C. (2000), "The effects of information ordering on investor perceptions: An experiment utilizing presidents´ letters", Journal of Financial and Strategic Decisions, Vol. 13 No. 3, pp. 71-81.Baker, H. E. and Kare, D. D. (1992), "Relationship between annual report readability and corporate financial performance", Management Research News, Vol. 15 No. 2, pp. 1-4.Barton, J. and Mercer, M. (2005), "To blame or not to blame: Analysts reactions to explanations of poor management performance", Journal of Accounting and Economics, Vol. 39 No. 3, pp. 509-533. 48
  • 51. Beattie, V.A. and Jones, M.J. (1992), "The use and abuse of graphs in annual reports: a theoretical framework and empirical study". Accounting and Business Research, Vol. 22 No. 88, pp. 291-303.Beattie, V. and Jones, M. J. (2002), "Measurement distortion of graphs in corporate reports: an experimental study", Accounting, Auditing & Accountability Journal, Vol. 15 No. 4, pp. 546-564.Beattie, V. A., McInnes, B. and Fearnley, S. (2004), "A methodology for analysing and evaluating narratives in annual reports: a comprehensive descriptive profile and metrics for disclosure quality attributes", Accounting Forum, Vol. 28 No., pp. 205-236.Bettman, J. R. and Weitz, B. A. (1983), "Attributions in the boardroom: Causal reasoning in corporate annual reports", Administrative Science Quarterly, Vol. 28 No. 2, pp. 165-183.Botosan, C.A. (1997), "Disclosure level and the cost of equity capital", The Accounting Review, Vol. 72, No. 3, pp. 323-349.Bowen, R. M., Davis, A. K. and Matsumoto, D. A. (2005), "Emphasis on pro forma versus GAAP earnings in quarterly press releases: determinants, SEC intervention and market reactions", The Accounting Review, Vol. 80 No. 4, pp. 1011-1038.Brennan, N. and Gray, S. J. (2000). Rhetoric and argument in financial reporting: disclosures in profit forecasts and takeover documents, ACCA Occasional Research Paper No. 31, Certified Accountants Educational Trust, London.Breton, G. and Taffler, R. J. (2001), "Accounting information and analyst stock recommendation decisions: a content analysis approach", Accounting and Business Research, Vol. 31 No. 2, pp. 91-101.Cassar, G. (2001), "Self-serving behaviour and the voluntary disclosure of capital market performance", Accounting Research Journal, Vol. 14 No. 2, pp. 126-137.Chow, C.W. and Wong-Boren, A. (1987), "Voluntary financial disclosure by Mexican corporations", The Accounting Review, Vol. 62 No. 3, pp. 533-541.Clatworthy, M. and Jones, M. J. (2001), "The effect of thematic structure on the variability of annual report readability", Accounting, Auditing & Accountability Journal, Vol. 14 No. 3, pp. 311-326. 49
  • 52. Clatworthy, M. and Jones, M. J. (2003), "Financial reporting of good news and bad news: evidence from accounting narratives", Accounting and Business Research, Vol. 33 No. 3, pp. 171-185.Clatworthy, M. and Jones, M. J. (2006), "Differential patterns of textual characteristics and company performance in the chairmans statement", Accounting, Auditing & Accountability Journal, Vol. 19 No. 4, pp. 493-511.Conway, M. (2007), "The subjective precision of computers: a methodological comparison with human coding in content analysis", Journalism and Mass Communication Quarterly, Vol. 83, No. 1, pp. 186-200.Cooke, T.E. (1989), "Disclosure in the corporate annual reports of Swedish companies", Accounting and Business Research, Vol. 19 No. 74, pp. 113-124.Courtis, J. K. (1986), "An investigation into annual report readability and corporate risk- return relationships", Accounting and Business Research, Vol. 16 No. 64, pp. 285- 294.Courtis, J. K. (1995), "Readability of annual reports: Western versus Asian evidence", Accounting, Auditing & Accountability Journal, Vol. 8 No. 2, pp. 4-17.Courtis, J. K. (1996), "Information redundancy in annual reports", Accountability and Performance, Vol. 12 No. 3, pp. 1-16.Courtis, J. K. (1998), "Annual report readability variability: tests of the obfuscation hypothesis", Accounting, Auditing & Accountability Journal, Vol. 11 No. 4, pp. 459-471.Courtis, J. K. (2004a), "Colour as visual rhetoric in financial reporting", Accounting Forum, Vol. 28 No. 3, pp. 265-281.Courtis, J. K. (2004b), "Corporate report obfuscation: Artefact or phenomenon?" British Accounting Review, Vol. 36 No. 3, pp. 291-312.Covaleski, M. A., Dirsmith, M. W. and Samuel, S. (1995), "The use of Accounting information in governmental regulation and public administration: The impact of John R. Commons and early institutional economists", Accounting Historians Journal, Vol. 22 No. 1, pp. 1-33.Davis, A.K., Piger, J.M. and Sedor, L.M. (2007), "Beyond the numbers: managers’ use of optimistic and pessimistic tone in earnings press releases". Working paper, December 2007, Available at Social Science Research Network (SSRN), 50
  • 53. Duriau, V.J., Reger, R.K. and Pfarrer, M.D. 2007, "A content analysis of the content analysis literature in organization studies: research themes, data sources, and methodological refinements ", Organizational Research Methods, Vol. 10 No. 1, pp. 5-34.Elliot, W. B. (2006), "Are investors influenced by pro forma emphasis and reconciliations in earnings announcements?" The Accounting Review, Vol. 81 No. 1, pp. 113-133.Frazier, K. B., Ingram, R. W. and Tennyson, B. M. (1984), "A methodology for the analysis of narrative accounting disclosures", Journal of Accounting Research, Vol. 22 No. 1, pp. 318-331.Gordon, E.A., Henry, E., Peytcheva, M. and Sun, L. (2007) Disclosure credibility and market reactions to restatements, Available at SSRN:, D. and Milne, M. J. (1996), "Some determinants of social and environmental disclosures in New Zealand companies", Accounting, Auditing & Accountability Journal, Vol. 9 No. 1, pp. 77-102.Henry, E. (2008). "Are investors influenced by how earnings press releases are written?", Journal of Business Communication 45 (4) 363-407.Hooghiemstra, R. (2001). "Cultural differences in self-serving behaviour in accounting narratives". APIRA Conference, Adelaide, 15-17 July 2001.Hoskin, R., Hughes, J. and Ricks, W. (1986), "Evidence on the incremental information content of additional firm disclosures made concurrently with earnings", Journal of Accounting Research, Vol. 24 Supplement, pp. 1-36.Ingram, R. W. and Frazier, K. B. (1980), "Environmental performance and corporate disclosure", Journal of Accounting Research, Vol. 18 No. 2, pp. 614-622.Ingram, R. W. and Frazier, K. B. (1983), "Narrative disclosures in annual reports", Journal of Business Research, Vol. 11 March, pp. 49-60.Jameson, D. (2000), "Telling the investment story: A narrative analysis of shareholder reports", The Journal of Business Communication, Vol. 37 No. 1, pp. 7-38.Johnson, W. B. and Schwartz, W. C. (2005), "Are investors misled by "pro forma" earnings?" Contemporary Accounting Research, Vol. 22 No. 4, pp. 915-963.Jones, M. J. (1988), "A longitudinal study of the readability of the chairmans narratives in the corporate reports of a UK company", Accounting and Business Research, Vol. 18 No. 72, pp. 297-305. 51
  • 54. Jones, M. J. (1997), "Critical appraisal of the cloze procedures use in the accounting domain", Accounting, Auditing & Accountability Journal, Vol. 10 No. 1, pp. 105- 128.Jones, M. J. and Shoemaker, P. A. (1994), "Accounting narratives: a review of empirical studies of content and readability", Journal of Accounting Literature, Vol. 13, pp. 142-184.Kelton, A. S. (2006). Internet financial reporting: The effects of hyperlinks and irrelevant information on investor judgements, Unpublished doctoral dissertation, The University of Tennessee, Knoxville.Kohut, G. F. and Segars, A. H. (1992), "The presidents letter to stockholders: and examination of corporate communication strategy", The Journal of Business Communication, Vol. 29 No. 1, pp. 7-21.Krippendorff, S. (1980), Content Analysis: An Introduction to its Methodology. Sage Publications, Beverly Hills, CA.Krische, S.D. (2005), "Investors evaluations of strategic prior-period benchmark disclosures in earnings announcements", The Accounting Review, Vol. 80 No. 1, pp. 243-268.Lang, M. and Lundholm, R. (2000), "Voluntary disclosure and equity offerings: reducing information asymmetry or hyping the stock?" Contemporary Accounting Research, Vol. 17 No. 4, pp. 623-662.Leary, M. R. and Kowalski, R. M. (1990), "Impression management: a literature review and two-component model", Psychological Bulletin, Vol. 107 No. 1, pp. 34-47.Lee, F., Peterson, C. and Tiedens, L. Z. (2004), "Mea culpa: predicting stock prices from organizational attributes", Personality and Social Psychology Bulletin, Vol. 30 No. 12, pp. 1636-1649.Lewellen, W. G., Park, T. and Ro, B. T. (1996), "Self-serving behavior in managers discretionary information disclosure decisions", Journal of Accounting and Economics, Vol. 21 No. 2, pp. 227-251.Lewis, N. R., Parker, L. D., Pound, G. D. and Sutcliffe, P. (1986), "Accounting report readability: The use of readability techniques", Accounting and Business Research, Vol. 16 Summer, pp. 199-213.Li, F. (2008), "Annual report readability, current earnings, and earnings persistence", Journal of Accounting and Economics, Vol. 45 No. 2-3, pp. 221-247. 52
  • 55. Linderman, A. (2001), "Computer content analysis and manual coding techniques: a comparative analysis", in Theory, Method and Practice in Computer Content Analysis, ed. West, M.D., Ablex, Westport CT, pp. 97-109.Llewellyn, S. (1999), "Narratives in accounting and management research", Accounting, Auditing and Accountability Journal, Vol. 12 No. 2, pp. 220-236.Lougee, B. A. and Marquardt, C. A. (2004), "Earnings informativeness and strategic disclosure: an empirical examination of pro forma net income", The Accounting Review, Vol. 79 No. 3, pp. 769-795.Maat, H. P. (2007), "How promotional language in press releases is dealt with by journalists", Journal of Business Communication, Vol. 44 No. 1, pp. 59-95.Matsumoto, D., Pronk, M. and Roelofsen, E. (2006), "Do analysts mitigate optimism by management?", Working paper, University of Washington.Merkl-Davies, D.M. (2007), "The obfuscation hypothesis re-examined: Analysing impression management in narrative annual report documents". Unpublished doctoral dissertation, Bangor University.Merkl-Davies, D. and Brennan, N. (2007). Discretionary disclosure strategies in corporate narratives: Incremental information or impression management?, Journal of Accounting Literature, Vol. 26, pp. 116-196.Milne, M. J. and Adler, R. A. (1999), "Exploring the reliability of social and environmental disclosures content analysis", Accounting, Auditing & Accountability Journal, Vol. 12 No. 2, pp. 237-256.Morris, R. (1994), "Computerized content analysis in management research: a demonstration of advantages and limitations", Journal of Management, Vol. 20, pp. 903-931.Neu, D. (1991), "Trust, impression management and the auditing profession", Critical Perspectives on Accounting, Vol. 2 No. 4, pp. 295-313.Neu, D., Warsame, H. A. and Pedwell, K. A. (1998), "Managing public impressions: Environmental disclosures in annual reports", Accounting, Organizations and Society, Vol. 23 No. 3, pp. 265-282.Ogden, S. and Clarke, J. (2005), "Customer disclosures, impression management and the construction of legitimacy: Corporate reports in the UK privatised water industry", Accounting, Auditing & Accountability Journal, Vol. 18 No. 3, pp. 313- 345. 53
  • 56. Oxford English Dictionary (1989). Oxford, Oxford University Press.Parker, L. D. (1982), "Corporate annual reporting: A mass communication perspective", Accounting and Business Research, Vol. 12 Autumn, pp. 279-286.Riahi-Belkaoui, A. (1999), "Disclosure adequacy and country risk", American Business Review, Vol. 17, No. 2, pp. 1-4.Riess, M., Rosenfeld, P., Melburg, V. and Tedeschi, J. T. (1981), "Self-serving attributions: biased private perceptions and distorted public descriptions", Journal of Personality and Social Psychology, Vol. 41 No. 2, pp. 224-231.Rutherford, B. (2003), "Obfuscation, textual complexity and the role of regulated narrative accounting disclosure in corporate governance", Journal of Management and Governance, Vol. 7 No. 2, pp. 187-210.Rutherford, B. (2005), "Genre analysis of corporate annual report narratives", Journal of Business Communication, Vol. 42 No. 4, pp. 349-378.Salancik, G. R. and Meindl, J. R. (1984), "Corporate attributions as strategic illusions of management control", Administrative Science Quarterly, Vol. 29, pp. 238-254.Schlenker, B. R. (1980), Impression Management: The Self-concept, Social Identity, and Interpersonal Relations, Belmont: Brooks-Cole, 1980.Schneider, D. J. (1981), Tactical self-presentations: Toward a broader conception. New York, Academic Press.Schrand, C. and Walther, B. R. (2000), "Strategic benchmarks in earnings announcements: The selective disclosure of prior-period earnings components", The Accounting Review, Vol. 75 No. 2, pp. 151-177.Short, J. C. and Palmer, T. B. (2003), "Organizational performance referents: an empirical examination of their content and influences", Organizational Behavior and Human Decision Processes, Vol. 90 No. 2, pp. 209-224.Skinner, D. J. (1994), "Why firms voluntary disclose bad news", Journal of Accounting Research, Vol. 32 No. 1, pp. 38-60.Smith, M. and Taffler, R. J. (1992a), "The chairmans statement and corporate financial performance", Accounting and Finance, Vol. 32 No. 2, pp. 75-90.Smith, M. and Taffler, R. J. (1992b), "Readability and understandability: Different measures of the textual complexity of accounting narrative", Accounting, Auditing & Accountability Journal, Vol. 5 No. 4, pp. 84-98. 54
  • 57. Smith, M. and Taffler, R. J. (1995), "The incremental effect of narrative accounting information in corporate annual reports", Journal of Business Finance and Accounting, Vol. 22 No. 8, pp. 1195-1210.Smith, M. and Taffler, R. J. (2000), "The chairmans statement: A content analysis of discretionary narrative disclosures", Accounting, Auditing & Accountability Journal, Vol. 13 No. 5, pp. 624-646.So, S. and Smith, M. (2002), "Colour graphics and text complexity in multivariate decision making", Accounting, Auditing & Accountability Journal, Vol. 15 No. 4, pp. 564-593.Staw, B. M., McKechnie, P. I. and Puffer, S. M. (1983), "The justification of organizational performance", Administrative Science Quarterly, Vol. 28 No. 4, pp. 582-600.Stevens, K., Stevens, K. C. and Stevens, W. P. (1992), "Measuring the readability of business writing: The Cloze procedure versus readability formulae", The Journal of Business Communication, Vol. 29 No. 4, pp. 367-382.Subramanian, R., Insley, R. G. and Blackwell, R. D. (1993), "Performance and readability: a comparison of annual reports of profitable and unprofitable corporations", The Journal of Business Communication, Vol. 30 No. 1, pp. 49-61.Sydserff, R. and Weetman, P. (1999), "A texture index for evaluating accounting narratives: An alternative to readability formulae", Accounting, Auditing & Accountability Journal, Vol. 12 No. 4, pp. 459-488.Sydserff, R. and Weetman, P. (2002), "Development in content analysis: a transitivity index and DICTION scores", Accounting, Auditing & Accountability Journal, Vol. 15 No. 4, pp. 523-545.Tennyson, B., Ingram, R. W. and Dugan, M. T. (1990), "Assessing the information content of narrative disclosures in explaining bankruptcy", Journal of Business Finance and Accounting, Vol. 17 No. 3, pp. 391-410.Tetlock, P.C., Saar-Tsechansky, M. and Macskassy, S. (2008), "More than words: Quantifying language to measure firms´ fundamentals", Journal of Finance, Vol. 63 No. 3, 1437-1467.Thomas, J. (1997), "Discourse in the marketplace: the making meaning of annual reports", Journal of Business Communication, Vol. 34 No. 1, pp. 47-66. 55
  • 58. Thompson, G. (1991), "Is accounting rhetorical? Methodology, Luca Pacioli and printing", Accounting, Organizations and Society, Vol. 16 No. 5/6, pp. 572-599.Wallace, W.A. (1992), "Whose power prevails in disclosure practices?", Auditing: A Journal of Practice & Theory, Vol. 11, Supplement, pp. 79-105.Warnock, K. (1992), "Structure and argument in accounting standards", Accounting and Business Research, Vol. 22 No. 86, pp. 179-188.Warnock, K. (2000), "Arguing about Goodwill", Irish Accounting Review, Vol. 7 No. 1, pp. 101-135.Weber, R. P. (1990), Basic Content Analysis. Sage University paper series on quantitative applications in the social sciences, No. 49. Newbury Park, CA.Yuthas, K., Rogers, R. and Dillard, J. F. (2002), "Communicative action and corporate annual reports", Journal of Business Ethics, Vol. 41 No. 1/2, pp. 141-157.Zarzeski, M.T, 1999 "Spontaneous harmonisation effects of culture and market forces on accounting disclosure practices", Accounting Horizons, Vol. 10 No. 1, pp. 18-37. 56
  • 59. Appendix 1: Selection of amounts from the P&L account (Wolseley plc Annual Report 2000)GROUP PROFIT & LOSS ACCOUNT – year ended 31 July 2000 2000 1999 £m £mTurnoverContinuing operations 5,782.6 5,245.3Acquisitions 429.7 - 6,212.3 5,245.3Discontinued activities 191.1 259.7 6,403.4 5,505.0 Increase overCosts less other income (6,030.2) (5,191.9) prior yearTrading profit before goodwill amortisation and exceptionals 385.7 330.2 +16.8%Exceptional items - (11.6) 385.7 318.6 +21.1%Goodwill amortisation (12.5) (5.5)Operating profitContinuing operations 338.1 290.5 +16.4%Acquisitions 20.3 - 358.4 290.5 +23.4%Discounted activities 14.8 22.6Operating profit 373.2 313.1 +19.2%Loss on disposal of operations (42.6) (3.1)Profit on ordinary activities before interest 330.6 310.0 +6.6%Net interest payable (28.3) (3.6)Profit on ordinary activities before tax 302.3 306.4 DecreaseTaxationOrdinary activities (114.4) (107.7)Exceptional credit 6.0 0.7 (108.4) (107.0)Profit after tax 193.9 199.4 DecreaseMonitory interests (0.4) (0.5)Profits for the year attributable to ordinary shareholders 193.5 198.9 DecreaseDividends (88.3) (78.9)Profits retained 105.2 120.0 DecreaseEarnings per shareBefore goodwill amortisation and exceptionals 42.26p 38.08p +11%Goodwill amortisation (2.17p) (0.96p)Exceptionals (6.38p) (2.43p)Basic earnings per share 33.71p 34.69p DecreaseDiluted earnings per share 33.67p 34.65p DecreaseKey: : This amount was selected for inclusion in the press release - : Numbers to identify the total possible profit/earnings per share amounts from which to select for disclosure in thepress release, ranked in order of size 57